def initialize(self, file_name, to_csv=False):
dismod_at.create_database(file_name, self.age_list, self.time_list,
self.integrand_table, self.node_table,
self.weight_table, self.covariate_table,
self.avgint_table, self.data_table,
self.prior_table, self.smooth_table, list(),
self.rate_table, self.mulcov_table,
self.option_table)
if to_csv:
flag = subprocess.call([program, file_name, 'init'])
if flag != 0:
sys.exit('command failed: flag = ' + str(flag))
dismod_at.db2csv_command(file_name)
def init_database(self, db2csv: bool = True):
dismod_at.create_database(self.path, self.age_list, self.time_list,
self.integrand_table, self.node_table,
self.weight_table, self.covariate_table,
self.avgint_table, self.data_table,
self.prior_table, self.smooth_table, list(),
self.rate_table, self.mulcov_table,
self.option_table)
if not os.path.exists(self.path):
os.mknod(self.path)
command = [program, self.path, 'init']
print(' '.join(command))
flag = subprocess.call(command)
if flag != 0:
sys.exit('The dismod_at init command failed')
if db2csv is True:
dismod_at.db2csv_command(self.path)
def example_db(file_name):
from math import log
def fun_iota_world(a, t):
value = 'prior_iota_world_value'
dage = 'prior_iota_world_smooth'
return ('prior_iota_world_value', None, None)
def fun_iota_child(a, t):
return ('prior_iota_child_value', None, None)
def fun_omega_world(a, t):
return (omega_true['world'], None, None)
def fun_omega_child_1(a, t):
return (omega_true['child_1'], None, None)
def fun_omega_child_2(a, t):
return (omega_true['child_2'], None, None)
# ----------------------------------------------------------------------
# age table:
age_list = [0.0, 100.0]
#
# time table:
time_list = [1990.0, 2200.0]
#
# integrand table:
integrand_table = [{'name': 'susceptible'}]
#
# node table:
node_table = [{
'name': 'world',
'parent': ''
}, {
'name': 'child_1',
'parent': 'world'
}, {
'name': 'child_2',
'parent': 'world'
}]
#
# weight table:
weight_table = list()
#
# covariate table:
covariate_table = list()
#
# mulcov table:
mulcov_table = list()
#
# avgint table: empty
avgint_table = list()
# ----------------------------------------------------------------------
# data table:
data_table = list()
# values that are the same for all data rows
row = {
'meas_value': 0.0, # not used (will be simulated)
'density': 'gaussian',
'weight': '',
'hold_out': False,
'time_lower': 2000.,
'time_upper': 2000.,
'integrand': 'susceptible',
'meas_std': 0.005,
'subgroup': 'world'
}
# values that change between rows:
node_name = ['world', 'child_1', 'child_2']
age_value = [10, 100, 10, 100]
time_value = [time_list[0], time_list[0], time_list[-1], time_list[-1]]
for data_id in range(n_data):
age = age_value[data_id % 4]
time = time_value[data_id % 4]
row['age_lower'] = age
row['age_upper'] = age
row['time_lower'] = time
row['time_upper'] = time
row['node'] = node_name[data_id % 3]
data_table.append(copy.copy(row))
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_iota_world_value
'name': 'prior_iota_world_value',
'density': 'uniform',
'lower': iota_true['world'] / 10.,
'upper': iota_true['world'] * 10.,
'mean': iota_true['world'] * 2.0,
},{ # prior_iota_child_value
'name': 'prior_iota_child_value',
'density': 'gaussian',
'mean': 0.0,
'std': 1.0
},{ # prior_iota_world_smooth
'name': 'prior_iota_world_smooth',
'density': 'gaussian',
'mean': 0.0,
'std': iota_true['world'] / 10.0
}
]
# ----------------------------------------------------------------------
# smooth table
smooth_table = [
{ # smooth_iota_world
'name': 'smooth_iota_world',
'age_id': [0, 1],
'time_id': [0, 1],
'fun': fun_iota_world
},
{ # smooth_iota_child
'name': 'smooth_iota_child',
'age_id': [0],
'time_id': [0],
'fun': fun_iota_child
},
{ # smooth_omega_world
'name': 'smooth_omega_world',
'age_id': [0, 1],
'time_id': [0, 1],
'fun': fun_omega_world
},
{ # smooth_omega_child_1
'name': 'smooth_omega_child_1',
'age_id': [0],
'time_id': [0],
'fun': fun_omega_child_1
},
{ # smooth_omega_child_2
'name': 'smooth_omega_child_2',
'age_id': [0],
'time_id': [0],
'fun': fun_omega_child_2
}
]
#
# nslist_table:
nslist_table = dict()
nslist_table['omega_nslist'] = [('child_1', 'smooth_omega_child_1'),
('child_2', 'smooth_omega_child_2')]
# ----------------------------------------------------------------------
# rate table:
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_iota_world',
'child_smooth': 'smooth_iota_child',
}, {
'name': 'omega',
'parent_smooth': 'smooth_omega_world',
'child_nslist': 'omega_nslist',
}]
# ----------------------------------------------------------------------
# option_table
option_table = [{
'name': 'rate_case',
'value': 'iota_pos_rho_zero'
}, {
'name': 'parent_node_name',
'value': 'world'
}, {
'name': 'ode_step_size',
'value': '10.0'
}, {
'name': 'random_seed',
'value': str(random_seed)
}, {
'name': 'quasi_fixed',
'value': 'true'
}, {
'name': 'max_num_iter_fixed',
'value': '50'
}, {
'name': 'print_level_fixed',
'value': '0'
}, {
'name': 'tolerance_fixed',
'value': '1e-8'
}, {
'name': 'max_num_iter_random',
'value': '50'
}, {
'name': 'print_level_random',
'value': '0'
}, {
'name': 'tolerance_random',
'value': '1e-10'
}, {
'name': 'zero_sum_child_rate',
'value': 'iota'
}]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
# ----------------------------------------------------------------------
return
def example_db(file_name, meas_noise_effect):
# ----------------------------------------------------------------------
# age table:
age_list = [0.0, 100.0]
#
# time table:
time_list = [1990.0, 2200.0]
#
# integrand table:
integrand_table = [{'name': 'Sincidence'}]
#
# node table:
node_table = [{'name': 'world', 'parent': ''}]
#
# weight table:
# The constant function 1.0, note any valid age and time id would work
name = 'constant'
fun = weight_fun
age_id = 0
time_id = 0
weight_table = [{
'name': name,
'age_id': [age_id],
'time_id': [time_id],
'fun': fun
}]
#
# covariate table:
covariate_table = [{'name': 'one', 'reference': 0.0}]
#
# mulcov table:
mulcov_table = [{
'covariate': 'one',
'type': 'meas_noise',
'effected': 'Sincidence',
'group': 'world',
'smooth': 'smooth_gamma'
}]
#
# avgint table: empty
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# data table:
data_table = list()
# values that are the same for all data rows
row = {
'meas_value': 0.0, # not used (will be simulated)
'density': 'gaussian',
'weight': 'constant',
'hold_out': False,
'time_lower': 2000.,
'time_upper': 2000.,
'integrand': 'Sincidence',
'meas_std': data_std,
'node': 'world',
'subgroup': 'world',
'one': 1.0
}
if meas_noise_effect == 'add_std_scale_all':
delta = data_std * (1.0 + gamma_true_scale)
gamma_true = gamma_true_scale
elif meas_noise_effect == 'add_std_scale_log':
delta = data_std * (1.0 + gamma_true_scale)
gamma_true = gamma_true_scale * data_std
elif meas_noise_effect == 'add_var_scale_all':
delta = data_std * math.sqrt(1.0 + gamma_true_scale)
gamma_true = gamma_true_scale
else:
assert meas_noise_effect == 'add_var_scale_log'
delta = data_std * math.sqrt(1.0 + gamma_true_scale)
gamma_true = gamma_true_scale * data_std * data_std
# values that change between rows:
for data_id in range(n_data):
row['meas_value'] = random.gauss(iota_true, delta)
fraction = data_id / float(n_data - 1)
age = age_list[0] + (age_list[-1] - age_list[0]) * fraction
row['age_lower'] = age
row['age_upper'] = age
data_table.append(copy.copy(row))
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_iota
'name': 'prior_iota',
'density': 'uniform',
'lower': iota_true / 100.0,
'upper': 1.0,
'mean': iota_true / 10.0
},
{ # prior_gamma
'name': 'prior_gamma',
'density': 'uniform',
'lower': 0.0,
'upper': 10.0 * gamma_true,
'mean': gamma_true / 10.0
}
]
# ----------------------------------------------------------------------
# smooth table
name = 'smooth_iota'
fun = fun_iota
age_id = 0
time_id = 0
smooth_table = [{
'name': name,
'age_id': [age_id],
'time_id': [time_id],
'fun': fun
}]
name = 'smooth_iota'
#
name = 'smooth_gamma'
fun = fun_gamma
smooth_table.append({
'name': name,
'age_id': [age_id],
'time_id': [time_id],
'fun': fun
})
# ----------------------------------------------------------------------
# rate table:
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_iota',
'child_smooth': None
}]
# ----------------------------------------------------------------------
# option_table
option_table = [{
'name': 'meas_noise_effect',
'value': meas_noise_effect
}, {
'name': 'rate_case',
'value': 'iota_pos_rho_zero'
}, {
'name': 'parent_node_name',
'value': 'world'
}, {
'name': 'random_seed',
'value': '0'
}, {
'name': 'zero_sum_child_rate',
'value': 'iota'
}, {
'name': 'quasi_fixed',
'value': 'false'
}, {
'name': 'derivative_test_fixed',
'value': 'second-order'
}, {
'name': 'max_num_iter_fixed',
'value': '100'
}, {
'name': 'print_level_fixed',
'value': '0'
}, {
'name': 'tolerance_fixed',
'value': '1e-10'
}]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
# ----------------------------------------------------------------------
return
def example_db(file_name):
def fun_rate_child(a, t):
return ('prior_rate_child', None, 'prior_child_diff')
def fun_rate_parent(a, t):
return ('prior_rate_parent', None, 'prior_parent_diff')
import dismod_at
# ----------------------------------------------------------------------
# age table
age_list = [0.0, 50.0, 100.0]
#
# time table
time_list = [1995.0, 2005.0, 2015.0]
#
# integrand table
integrand_table = [{'name': 'Sincidence'}]
#
# node table: world -> north_america
# north_america -> (united_states, canada)
node_table = [{
'name': 'world',
'parent': ''
}, {
'name': 'north_america',
'parent': 'world'
}, {
'name': 'united_states',
'parent': 'north_america'
}, {
'name': 'canada',
'parent': 'north_america'
}]
#
# weight table:
weight_table = list()
#
# covariate table: no covriates
covariate_table = list()
#
# mulcov table
mulcov_table = list()
#
# avgint table:
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# data table:
data_table = list()
# write out data
row = {
'density': 'gaussian',
'weight': '',
'hold_out': False,
'time_lower': 2000.0,
'time_upper': 2000.0,
'age_lower': 50.0,
'age_upper': 50.0,
'integrand': 'Sincidence',
'subgroup': 'world',
}
row['node'] = 'north_america'
row['meas_value'] = iota_parent_true
row['meas_std'] = row['meas_value'] * 1e-1
data_table.append(copy.copy(row))
#
row['node'] = 'united_states'
row['meas_value'] = iota_parent_true * exp(united_states_random_effect)
row['meas_std'] = row['meas_value'] * 1e-1
data_table.append(copy.copy(row))
#
row['node'] = 'canada'
row['meas_value'] = iota_parent_true * exp(-united_states_random_effect)
row['meas_std'] = row['meas_value'] * 1e-1
data_table.append(copy.copy(row))
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_rate_parent
'name': 'prior_rate_parent',
'density': 'uniform',
'lower': iota_parent_true / 100.,
'upper': iota_parent_true * 100.,
# set prior so north_amaerica is set to value for united_states
'mean': iota_parent_true * exp( united_states_random_effect )
},{ # prior_rate_child
'name': 'prior_rate_child',
'density': 'gaussian',
'mean': 0.0,
# std is very large so like a uniform distribution
'std': 100.0,
},{ # prior_parent_diff
'name': 'prior_parent_diff',
'density': 'log_gaussian',
'mean': 0.0,
'std': 0.1,
'eta': 1e-8
},{ # prior_child_diff
'name': 'prior_child_diff',
'density': 'gaussian',
'mean': 0.0,
'std': 0.1,
}
]
# ----------------------------------------------------------------------
# smooth table
last_time_id = 2
smooth_table = [
{ # smooth_rate_child
'name': 'smooth_rate_child',
'age_id': [0],
'time_id': [0, last_time_id],
'fun': fun_rate_child
},
{ # smooth_rate_parent
'name': 'smooth_rate_parent',
'age_id': [0],
'time_id': [0, last_time_id],
'fun': fun_rate_parent
}
]
# ----------------------------------------------------------------------
# rate table
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_rate_parent',
'child_smooth': 'smooth_rate_child',
}]
# ----------------------------------------------------------------------
# option_table
option_table = [{
'name': 'parent_node_name',
'value': 'north_america'
}, {
'name': 'derivative_test_random',
'value': 'second-order'
}, {
'name': 'max_num_iter_random',
'value': '100'
}, {
'name': 'tolerance_random',
'value': '1e-10'
}, {
'name': 'method_random',
'value': 'ipopt_solve'
}]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
# ----------------------------------------------------------------------
return
def example_db(file_name):
def fun_iota_parent(a, t):
return ('prior_iota_parent_value', None, None)
def fun_iota_child(a, t):
return ('prior_iota_child', None, None)
def fun_iota_group(a, t):
return ('prior_iota_group', None, None)
def fun_iota_subgroup(a, t):
return ('prior_iota_subgroup', None, None)
#
# age_table
age_list = [0.0, 100.0]
#
# time_table
time_list = [1990.0, 2020.0]
#
# node_table
node_table = [
{
'name': 'p1',
'parent': ''
},
{
'name': 'c1',
'parent': 'p1'
},
{
'name': 'c2',
'parent': 'p1'
},
]
#
# subgroup_table
subgroup_table = [
{
'subgroup': 'none',
'group': 'none'
},
{
'subgroup': 's1',
'group': 'g1'
},
{
'subgroup': 's2',
'group': 'g1'
},
]
# rate_table
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_iota_parent',
'child_smooth': 'smooth_iota_child',
}]
# covariate_table
covariate_table = [{'name': 'one', 'reference': 0.0}]
# mulcov_table
mulcov_table = [{
'covariate': 'one',
'type': 'rate_value',
'effected': 'iota',
'group': 'g1',
'smooth': 'smooth_iota_group',
'subsmooth': 'smooth_iota_subgroup',
}]
# prior_table
prior_table = [
{ # prior_iota_parent_value
'name': 'prior_iota_parent_value',
'density': 'uniform',
'lower': iota_parent_true / 100.0,
'upper': iota_parent_true * 100.0,
'mean': iota_parent_true / 2.0,
},{ # prior_iota_child
'name': 'prior_iota_child',
'density': 'uniform',
'mean': 1e-3,
},{ # prior_iota_group
'name': 'prior_iota_group',
'density': 'uniform',
'mean': 0.0,
},{ # prior_iota_subgroup
'name': 'prior_iota_subgroup',
'density': 'uniform',
'mean': 1e-3,
}
]
# smooth_table
smooth_table = [
{ # smooth_iota_parent
'name': 'smooth_iota_parent',
'age_id': [0],
'time_id': [0],
'fun': fun_iota_parent
},
{ # smooth_iota_child
'name': 'smooth_iota_child',
'age_id': [0],
'time_id': [0],
'fun': fun_iota_child
},
{ # smooth_iota_group
'name': 'smooth_iota_group',
'age_id': [0],
'time_id': [0],
'fun': fun_iota_group
},
{ # smooth_iota_subgroup
'name': 'smooth_iota_subgroup',
'age_id': [0],
'time_id': [0],
'fun': fun_iota_subgroup
}
]
# weight table:
weight_table = list()
# nslist_table
nslist_table = dict()
# option_table
option_table = [
{
'name': 'parent_node_name',
'value': 'p1'
},
{
'name': 'rate_case',
'value': 'iota_pos_rho_zero'
},
{
'name': 'zero_sum_child_rate',
'value': 'iota'
},
{
'name': 'zero_sum_mulcov_group',
'value': 'g1'
},
{
'name': 'quasi_fixed',
'value': 'false'
},
{
'name': 'max_num_iter_fixed',
'value': '100'
},
{
'name': 'print_level_fixed',
'value': '0'
},
{
'name': 'tolerance_fixed',
'value': '1e-12'
},
{
'name': 'derivative_test_fixed',
'value': 'adaptive'
},
]
# integrand_table
integrand_table = [{'name': 'Sincidence'}]
# ------------------------------------------------------------------------
# data_table
data_table = list()
# values that are the same for all data points
row = {
'integrand': 'Sincidence',
'density': 'gaussian',
'weight': '',
'hold_out': False,
'age_lower': 50.0,
'age_upper': 50.0,
'time_lower': 2000.0,
'time_upper': 2000.0,
'one': 1.0,
}
for child in ['c1', 'c2']:
for subgroup in ['s1', 's2', 'none']:
# true value of iota for this child, subgroup
iota = avg_integrand(child, subgroup)
# measurement standard deviation is used during fittting
meas_std = iota * meas_cv
# for this example, measurements are simulated without any noise
meas_value = iota
#
row['node'] = child
row['subgroup'] = subgroup
row['meas_value'] = meas_value
row['meas_std'] = meas_std
data_table.append(copy.copy(row))
# ------------------------------------------------------------------------
# avgint_table
avgint_table = list()
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
def example_db(file_name):
def fun_meas_subgroup(a, t):
return ('prior_meas_subgroup', None, 'prior_gauss_diff')
def fun_rate_parent(a, t):
return ('prior_rate_parent', None, 'prior_gauss_diff')
import dismod_at
# ----------------------------------------------------------------------
# age list
age_list = [0.0, 50.0, 100.0]
#
# time list
time_list = [1990.0, 2010.0]
#
# integrand table
integrand_table = [{'name': 'Sincidence'}, {'name': 'remission'}]
#
# node table: north_america -> (united_states, canada)
node_table = [{
'name': 'north_america',
'parent': ''
}, {
'name': 'united_states',
'parent': 'north_america'
}, {
'name': 'canada',
'parent': 'north_america'
}]
#
# subgroup_table
subgroup_table = [
{
'subgroup': 'united_states',
'group': 'north_america'
},
{
'subgroup': 'canada',
'group': 'north_america'
},
]
#
# mulcov table
mulcov_table = [
{ # subgroup covariate multiplers effecting Sincidence
'covariate': 'one',
'type': 'meas_value',
'effected': 'Sincidence',
'group': 'north_america',
'smooth': None,
'subsmooth': 'smooth_meas_subgroup'
},
{ # subgroup covariate multipliers effecting remissions
'covariate': 'one',
'type': 'meas_value',
'effected': 'remission',
'group': 'north_america',
'smooth': None,
'subsmooth': 'smooth_meas_subgroup'
}
]
#
# weight table:
weight_table = list()
#
# covariate table: no covriates
covariate_table = [{
'name': 'one',
'reference': 0.0,
'max_difference': None
}]
#
# avgint table: same order as list of integrands
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# data table
data_table = list()
# write out data
row = {
'density': 'gaussian',
'weight': '',
'hold_out': False,
'age_lower': 50.0,
'age_upper': 50.0,
'one': 1.0,
}
for data_id in range(number_data):
if data_id % 2 == 0:
row['node'] = 'united_states'
row['subgroup'] = 'united_states'
row['data_name'] = 'us_' + str(data_id / 2)
effect_true = -subgroup_mulcov
if data_id % 3 == 2:
row['node'] = 'canada'
row['subgroup'] = 'canada'
row['data_name'] = 'ca_' + str(data_id / 2)
effect_true = +subgroup_mulcov
if data_id % 4 < 2:
row['time_lower'] = 1990.0
row['time_upper'] = 1990.0
else:
row['time_lower'] = 2010.0
row['time_upper'] = 2010.0
#
if data_id < number_data / 2:
rate_true = iota_parent
row['integrand'] = 'Sincidence'
else:
rate_true = rho_parent
row['integrand'] = 'remission'
#
meas_mean = math.exp(effect_true) * rate_true
meas_std = meas_mean * measurement_cv
row['meas_value'] = random.gauss(meas_mean, meas_std)
row['meas_std'] = meas_std
data_table.append(copy.copy(row))
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_rate_parent
'name': 'prior_rate_parent',
'density': 'uniform',
'lower': min(iota_parent, rho_parent) / 100.0,
'upper': max(iota_parent, rho_parent) * 100.0,
'mean': 2.0 * max(iota_parent, rho_parent),
},{ # prior_meas_subgroup
'name': 'prior_meas_subgroup',
'density': 'gaussian',
'mean': 0.0,
'std': 100.0, # very large so like uniform distribution
},{ # prior_gauss_diff
'name': 'prior_gauss_diff',
'density': 'uniform',
'mean': 0.0,
'std': 100.0, # very large so like uniform distribution
}
]
# ----------------------------------------------------------------------
# smooth table
smooth_table = [
{ # smooth_meas_subgroup
'name': 'smooth_meas_subgroup',
'age_id': [0],
'time_id': [0, 1],
'fun': fun_meas_subgroup
},
{ # smooth_rate_parent
'name': 'smooth_rate_parent',
'age_id': [0],
'time_id': [0, 1],
'fun': fun_rate_parent
}
]
# ----------------------------------------------------------------------
# rate table
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_rate_parent',
'child_smooth': None,
}, {
'name': 'rho',
'parent_smooth': 'smooth_rate_parent',
'child_smooth': None,
}]
# ----------------------------------------------------------------------
# option_table
option_table = [{
'name': 'parent_node_name',
'value': 'north_america'
}, {
'name': 'zero_sum_mulcov_group',
'value': 'north_america'
}, {
'name': 'random_seed',
'value': '0'
}, {
'name': 'ode_step_size',
'value': '10.0'
}, {
'name': 'rate_case',
'value': 'iota_pos_rho_pos'
}, {
'name': 'quasi_fixed',
'value': 'true'
}, {
'name': 'derivative_test_fixed',
'value': 'first-order'
}, {
'name': 'max_num_iter_fixed',
'value': '100'
}, {
'name': 'print_level_fixed',
'value': '0'
}, {
'name': 'tolerance_fixed',
'value': '1e-7'
}, {
'name': 'derivative_test_random',
'value': 'second-order'
}, {
'name': 'max_num_iter_random',
'value': '100'
}, {
'name': 'print_level_random',
'value': '0'
}, {
'name': 'tolerance_random',
'value': '1e-10'
}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
def create_database():
import dismod_at
import copy
#
# file_name, age_list, time_list, integrand_table
file_name = 'example.db'
age_list = [50.0]
time_list = [2000.0]
integrand_table = [{'name': 'mtother', 'minimum_meas_cv': 0.0}]
#
# node_table
node_table = [{
'name': 'world',
'parent': ''
}, {
'name': 'north_america',
'parent': 'world'
}, {
'name': 'united_states',
'parent': 'north_america'
}, {
'name': 'canada',
'parent': 'north_america'
}]
# weight_table
fun = constant_weight_fun
weight_table = [{
'name': 'constant',
'age_id': [0],
'time_id': [0],
'fun': fun
}]
# covariate_table
covariate_table = [{
'name': 'sex',
'reference': 0.0,
'max_difference': 0.6
}]
# data_table
data_table = []
row = {
'integrand': 'mtother',
'density': 'log_gaussian',
'weight': 'constant',
'hold_out': False,
'meas_std': 1e-5,
'age_lower': 0.0,
'age_upper': 100.0,
'time_lower': 1990.0,
'time_upper': 2010.0,
'sex': 0.5,
'subgroup': 'world',
}
row['data_name'] = 'one'
row['node'] = 'north_america'
row['meas_value'] = 1.0e-5
data_table.append(copy.copy(row))
row['data_name'] = 'two'
row['node'] = 'united_states'
row['meas_value'] = 1.5e-5
data_table.append(copy.copy(row))
row['data_name'] = 'three'
row['node'] = 'canada'
row['meas_value'] = 0.5e-5
data_table.append(copy.copy(row))
#
# prior_table
prior_table = [{
'name': 'zero',
'density': 'uniform',
'lower': 0.0,
'upper': 0.0,
'mean': 0.0,
}, {
'name': 'one',
'density': 'uniform',
'lower': 1.0,
'upper': 1.0,
'mean': 1.0,
}, {
'name': 'uniform_01',
'density': 'uniform',
'lower': 0.0,
'upper': 1.0,
'mean': 0.1,
}, {
'name': 'gaussian_01',
'density': 'gaussian',
'mean': 0.0,
'std': 1.0,
}, {
'name': 'log_gaussian',
'density': 'log_gaussian',
'mean': 0.0,
'eta': 1e-6
}]
#
# smooth list
smooth_table = [{
'name': 'uniform_01_constant',
'age_id': [0],
'time_id': [0],
'mulstd_value_prior_name': None,
'mulstd_dage_prior_name': None,
'mulstd_dtime_prior_name': None,
'fun': smooth_uniform_01_fun
}, {
'name': 'gaussian_01_constant',
'age_id': [0],
'time_id': [0],
'mulstd_value_prior_name': None,
'mulstd_dage_prior_name': None,
'mulstd_dtime_prior_name': None,
'fun': smooth_gaussian_01_fun
}]
#
# rate_table
rate_table = [{
'name': 'pini',
'parent_smooth': 'uniform_01_constant',
'child_smooth': 'gaussian_01_constant',
'child_nslist': None
}, {
'name': 'iota',
'parent_smooth': 'uniform_01_constant',
'child_smooth': 'gaussian_01_constant',
'child_nslist': None
}, {
'name': 'rho',
'parent_smooth': 'uniform_01_constant',
'child_smooth': 'gaussian_01_constant',
'child_nslist': None
}, {
'name': 'chi',
'parent_smooth': 'uniform_01_constant',
'child_smooth': 'gaussian_01_constant',
'child_nslist': None
}, {
'name': 'omega',
'parent_smooth': 'uniform_01_constant',
'child_smooth': 'gaussian_01_constant',
'child_nslist': None
}]
#
# mulcov_table
mulcov_table = [{
'covariate': 'sex',
'type': 'rate_value',
'effected': 'omega',
'group': 'world',
'smooth': 'uniform_01_constant'
}]
#
# option_table
option_table = [{
'name': 'parent_node_name',
'value': 'world'
}, {
'name': 'ode_step_size',
'value': '10.0'
}, {
'name': 'random_seed',
'value': '0'
}, {
'name': 'rate_case',
'value': 'iota_pos_rho_pos'
}, {
'name': 'tolerance',
'value': '1e-8'
}, {
'name': 'max_num_iter',
'value': '100'
}, {
'name': 'print_level',
'value': '0'
}, {
'name': 'derivative_test',
'value': 'second-order'
}]
# avgint_table
avgint_table = []
row = {
'integrand': 'mtother',
'weight': 'constant',
'age_lower': 0.0,
'age_upper': 100.0,
'time_lower': 1990.0,
'time_upper': 2010.0,
'sex': 0.5,
'subgroup': 'world',
}
row['node'] = 'north_america'
avgint_table.append(copy.copy(row))
row['node'] = 'united_states'
avgint_table.append(copy.copy(row))
row['node'] = 'canada'
row['meas_value'] = 0.5e-5
avgint_table.append(copy.copy(row))
# nslist_table
nslist_table = dict()
#
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
# ----------------------------------------------------------------------
# Check database
# ----------------------------------------------------------------------
#
# connection
new = False
connection = dismod_at.create_connection(file_name, new)
#
# age_table
tbl_name = 'age'
col_name = ['age']
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
col_list = column_as_list(row_list, 0)
assert col_list == age_list
#
# time_table
tbl_name = 'time'
col_name = ['time']
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
col_list = column_as_list(row_list, 0)
assert col_list == time_list
#
# intergrand_table
tbl_name = 'integrand'
col_name = ['integrand_name']
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [['mtother']]
assert row_list == check_list
#
# weight_table
tbl_name = 'weight'
col_name = ['weight_name', 'n_age', 'n_time']
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [['constant', 1, 1]]
assert row_list == check_list
#
# weight_grid_table
tbl_name = 'weight_grid'
col_name = ['weight_id', 'age_id', 'time_id', 'weight']
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [[0, 0, 0, 1.0]]
assert row_list == check_list
#
# covariate_table
tbl_name = 'covariate'
col_name = ['covariate_name', 'reference']
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [['sex', 0.0]]
assert row_list == check_list
#
# node_table
tbl_name = 'node'
col_name = ['node_name', 'parent']
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [['world', None], ['north_america', 0], ['united_states', 1],
['canada', 1]]
assert row_list == check_list
#
# data_table
tbl_name = 'data'
col_name = ['integrand_id', 'node_id', 'meas_value', 'meas_std']
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [[0, 1, 1.0e-5, 1e-5], [0, 2, 1.5e-5, 1e-5],
[0, 3, 0.5e-5, 1e-5]]
assert row_list == check_list
#
# prior_table
tbl_name = 'prior'
col_name = ['prior_name', 'density_id', 'lower', 'upper']
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [['zero', 0, 0.0, 0.0], ['one', 0, 1.0, 1.0],
['uniform_01', 0, 0.0, 1.0], ['gaussian_01', 1, None, None],
['log_gaussian', 4, None, None]]
assert row_list == check_list
#
# smooth_table
tbl_name = 'smooth'
col_name = [
'smooth_name', 'n_age', 'n_time', 'mulstd_value_prior_id',
'mulstd_dage_prior_id', 'mulstd_dtime_prior_id'
]
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [['uniform_01_constant', 1, 1, None, None, None],
['gaussian_01_constant', 1, 1, None, None, None]]
assert row_list == check_list
#
# smooth_grid_table
tbl_name = 'smooth_grid'
col_name = [
'smooth_id', 'age_id', 'time_id', 'value_prior_id', 'dage_prior_id',
'dtime_prior_id'
]
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [[0, 0, 0, 2, None, None], [1, 0, 0, 3, None, None]]
assert row_list == check_list
#
# rate_table
tbl_name = 'rate'
col_name = ['rate_name', 'parent_smooth_id', 'child_smooth_id']
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [['pini', 0, 1], ['iota', 0, 1], ['rho', 0, 1], ['chi', 0, 1],
['omega', 0, 1]]
assert row_list == check_list
#
# mulcov_table
tbl_name = 'mulcov'
col_name = [
'mulcov_type', 'rate_id', 'integrand_id', 'covariate_id', 'group_id',
'subgroup_smooth_id', 'group_smooth_id'
]
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [['rate_value', 4, None, 0, 0, None, 0]]
assert row_list == check_list
#
# option_table
tbl_name = 'option'
col_name = ['option_name', 'option_value']
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [['parent_node_name', 'world'], ['ode_step_size', '10.0'],
['random_seed', '0'], ['rate_case', 'iota_pos_rho_pos'],
['tolerance', '1e-8'], ['max_num_iter', '100'],
['print_level', '0'], ['derivative_test', 'second-order']]
assert row_list == check_list
#
# avgint_table
tbl_name = 'avgint'
col_name = ['integrand_id', 'node_id', 'age_lower', 'age_upper']
row_list = dismod_at.get_row_list(connection, tbl_name, col_name)
check_list = [[0, 1, 0.0, 100.0], [0, 2, 0.0, 100.0], [0, 3, 0.0, 100.0]]
assert row_list == check_list
# ----------------------------------------------------------------------
connection.close()
print('create_database: OK')
def example_db (file_name) :
# ----------------------------------------------------------------------
# age table:
age_list = [ 0.0, 5.0, 15.0, 35.0, 50.0, 75.0, 90.0, 100.0 ]
#
# time table:
time_list = [ 1990.0, 2000.0, 2010.0, 2200.0 ]
#
# integrand table:
integrand_table = [
{ 'name':'prevalence', 'minimum_meas_cv':0.0 }
]
#
# node table:
node_table = [ { 'name':'world', 'parent':'' } ]
for i in range(n_children) :
name = 'child_' + str(i + 1)
node_table.append( { 'name':name, 'parent':'world' } )
#
# weight table:
weight_table = list()
#
# covariate table:
covariate_table = [
{'name':'income', 'reference':0.5, 'max_difference':None}
]
#
# mulcov table:
mulcov_table = [
{
'covariate': 'income',
'type': 'rate_value',
'effected': 'iota',
'group': 'world',
'smooth': 'smooth_mulcov'
}
]
# ----------------------------------------------------------------------
# data table:
data_table = list()
#
density_list = [ "gaussian", "students", "log_gaussian", "log_students" ]
# values that are the same for all data rows
row = {
'meas_value': 1., # measurement is way off truth and not used
'eta': 1e-6,
'nu': 10,
'weight': '',
'hold_out': False,
'time_lower': 2000.,
'time_upper': 2000.,
'subgroup': 'world',
}
# values that change between rows:
for data_id in range( n_data ) :
density = density_list[ data_id % 4 ]
row['density'] = density
#
# one percent coefficient of variation
if density == 'gaussian' or density == 'students' :
row['meas_std'] = iota_parent_true / 100.
else :
row['meas_std'] = row['meas_value'] / 100.
fraction = data_id / float(n_data-1)
age = age_list[0] + (age_list[-1] - age_list[0])*fraction
row['age_lower'] = age
row['age_upper'] = age
row['node'] = 'child_' + str( (data_id % n_children) + 1 )
row['income'] = fraction
row['integrand'] = integrand_table[0]['name']
data_table.append( copy.copy(row) )
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_zero
'name': 'prior_zero',
'density': 'uniform',
'lower': 0.0,
'upper': 0.0,
'mean': 0.0,
'std': None,
'eta': None
}, { # prior_none
'name': 'prior_none',
'density': 'uniform',
'lower': None,
'upper': None,
'mean': 0.0,
'std': None,
'eta': None
},{ # prior_gauss_zero
'name': 'prior_gauss_zero',
'density': 'gaussian',
'lower': None,
'upper': None,
'mean': 0.0,
'std': 0.01,
'eta': None
},{ # prior_loggauss_zero
'name': 'prior_loggauss_zero',
'density': 'log_gaussian',
'lower': None,
'upper': None,
'mean': 0.0,
'std': 0.1,
'eta': 1e-6
},{ # prior_iota_parent
'name': 'prior_iota_parent',
'density': 'uniform',
'lower': iota_parent_true / 100.,
'upper': iota_parent_true * 10.0,
'mean': iota_parent_true / 2.0,
'std': None,
'eta': None
},{ # prior_mulcov
'name': 'prior_mulcov',
'density': 'uniform',
'lower': -2.0,
'upper': +2.0,
'mean': 0.0,
'std': None,
'eta': None
}
]
# ----------------------------------------------------------------------
# smooth table
name = 'smooth_rate_child'
fun = fun_rate_child
age_id = int( len( age_list ) / 2 )
time_id = int( len( time_list ) / 2 )
smooth_table = [
{'name':name, 'age_id':[age_id], 'time_id':[time_id], 'fun':fun }
]
name = 'smooth_iota_parent'
fun = fun_iota_parent
smooth_table.append(
{'name':name, 'age_id':[age_id], 'time_id':[time_id], 'fun':fun }
)
name = 'smooth_mulcov'
fun = fun_mulcov
smooth_table.append(
{'name':name, 'age_id':[age_id], 'time_id':[time_id], 'fun':fun }
)
# no standard deviation multipliers
for dictionary in smooth_table :
for name in [ 'value' , 'dage', 'dtime' ] :
key = 'mulstd_' + name + '_prior_name'
value = None
dictionary[key] = value
# ----------------------------------------------------------------------
# rate table:
rate_table = [
{ 'name': 'iota',
'parent_smooth': 'smooth_iota_parent',
'child_smooth': 'smooth_rate_child',
'child_nslist': None
},
]
# ----------------------------------------------------------------------
# option_table
option_table = [
{ 'name':'rate_case', 'value':'iota_pos_rho_zero' },
{ 'name':'parent_node_name', 'value':'world' },
{ 'name':'ode_step_size', 'value':'10.0' },
{ 'name':'random_seed', 'value':random_seed_str },
{ 'name':'zero_sum_child_rate', 'value':'iota' },
{ 'name':'quasi_fixed', 'value':'true' },
{ 'name':'derivative_test_fixed', 'value':'none' },
{ 'name':'max_num_iter_fixed', 'value':'100' },
{ 'name':'print_level_fixed', 'value':'0' },
{ 'name':'tolerance_fixed', 'value':'1e-6' },
{ 'name':'derivative_test_random', 'value':'none' },
{ 'name':'max_num_iter_random', 'value':'100' },
{ 'name':'print_level_random', 'value':'0' },
{ 'name':'tolerance_random', 'value':'1e-8' },
{ 'name':'method_random', 'value':'ipopt_random' }
]
# ----------------------------------------------------------------------
# avgint table: empty
avgint_table = list()
# ----------------------------------------------------------------------
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [ { 'subgroup':'world', 'group':'world' } ]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(
file_name,
age_list,
time_list,
integrand_table,
node_table,
subgroup_table,
weight_table,
covariate_table,
avgint_table,
data_table,
prior_table,
smooth_table,
nslist_table,
rate_table,
mulcov_table,
option_table
)
# ----------------------------------------------------------------------
return
def example_db (file_name) :
def fun_rate_parent(a, t) :
return ('prior_iota_parent', None, None)
import dismod_at
# ----------------------------------------------------------------------
# age table
age_list = [ 0.0, 50.0, 100.0 ]
#
# time table
time_list = [ 1995.0, 2005.0, 2015.0 ]
#
# integrand table
integrand_table = [
{ 'name':'Sincidence' }
]
#
# node table: north_america -> (united_states, canada)
node_table = [
{ 'name':'north_america', 'parent':'' },
{ 'name':'united_states', 'parent':'north_america' },
{ 'name':'canada', 'parent':'north_america' }
]
#
# weight table:
weight_table = list()
#
# covariate table: no covriates
covariate_table = list()
#
# mulcov table
mulcov_table = list()
#
# avgint table: same order as list of integrands
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# data table: same order as list of integrands
data_table = list()
meas_std = meas_cv * iota_true
row = {
'density': 'gaussian',
'weight': '',
'hold_out': False,
'time_lower': 2000.0,
'time_upper': 2000.0,
'age_lower': 50.0,
'age_upper': 50.0,
'integrand': 'Sincidence',
'meas_std': meas_std,
'nu': nu
}
random.seed(random_seed)
for data_id in range( n_data ):
row['node'] = 'north_america'
meas_value = random.gauss(iota_true, meas_std)
#
if data_id < n_outlier :
meas_value = random.gauss( 10.0 * iota_true, 2.0 * iota_true )
row['meas_value'] = meas_value
data_table.append( copy.copy(row) )
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_iota_parent
'name': 'prior_iota_parent',
'density': 'uniform',
'lower': 1e-4,
'upper': 1.0,
'mean': 0.1,
}
]
# ----------------------------------------------------------------------
# smooth table
smooth_table = [
{ # smooth_rate_parent
'name': 'smooth_rate_parent',
'age_id': [ 0 ],
'time_id': [ 0 ],
'fun': fun_rate_parent
}
]
# ----------------------------------------------------------------------
# rate table
rate_table = [
{
'name': 'iota',
'parent_smooth': 'smooth_rate_parent',
}
]
# ----------------------------------------------------------------------
# option_table
option_table = [
{ 'name':'parent_node_name', 'value':'north_america' },
{ 'name':'ode_step_size', 'value':'10.0' },
{ 'name':'rate_case', 'value':'iota_pos_rho_zero' },
{ 'name':'quasi_fixed', 'value':'false' },
{ 'name':'max_num_iter_fixed', 'value':'100' },
{ 'name':'print_level_fixed', 'value':'0' },
{ 'name':'tolerance_fixed', 'value':'1e-11' },
]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [ { 'subgroup':'world', 'group':'world' } ]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(
file_name,
age_list,
time_list,
integrand_table,
node_table,
subgroup_table,
weight_table,
covariate_table,
avgint_table,
data_table,
prior_table,
smooth_table,
nslist_table,
rate_table,
mulcov_table,
option_table
)
def get_started_db():
import dismod_at
from math import exp
# ----------------------------------------------------------------------
# age list
age_list = [0.0, 100.0]
#
# time list
time_list = [1995.0, 2015.0]
#
# only one integrand in this example
integrand_table = [{'name': 'susceptible'}]
#
# just the world (which has no parent)
node_table = [{'name': 'world', 'parent': ''}]
#
# weight table is represented by one (age, time) point with value 1.0
fun = constant_one_fun
weight_table = [{
'name': 'constant_one',
'age_id': [0],
'time_id': [0],
'fun': fun
}]
#
# covariate table: the reference value for income is zero
covariate_table = [{'name': 'income', 'reference': 0.0}]
# ---------------------------------------------------------------------
# avgint table: predict the susceptible fraction for no income at age 100
avgint_table = [{
'avgint_info': 'a0',
'integrand': 'susceptible',
'node': 'world',
'subgroup': 'world',
'weight': 'constant_one',
'age_lower': 100.0,
'age_upper': 100.0,
'time_lower': 2000.0,
'time_upper': 2000.0,
'income': 0.0,
}]
# ---------------------------------------------------------------------
# data table: measure the susceptible fraction for income 1000 at age 50
# (no noise in this simulated data, but modeled as having noise)
adjusted_omega = unknown_omega_world * exp(
known_income_multiplier * 1000.0)
meas_value = exp(-adjusted_omega * 50.0)
meas_std = meas_value / 20.
data_table = [{
'c_data_info': 'd1',
'integrand': 'susceptible',
'node': 'world',
'subgroup': 'world',
'weight': 'constant_one',
'age_lower': 50.0,
'age_upper': 50.0,
'time_lower': 2000.0,
'time_upper': 2000.0,
'income': 1000.0,
'density': 'gaussian',
'hold_out': False,
'meas_value': meas_value,
'meas_std': meas_std,
}]
# ---------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_not_used
# (not used because there are no age or time differences)
'name': 'prior_not_used',
'density': 'uniform',
'mean': 0.0,
},{ # prior_omega_parent
# omega for the parent is constant in (age, time) and
# and is a uniform distribution on the interval [1e-4, 1.0].
# Becasue the prior is uniform, its mean 1e-1 is only used to
# start the optimization
'name': 'prior_omega_parent',
'density': 'uniform',
'lower': 1e-4,
'upper': 1.0,
'mean': 1e-1,
},{ # prior_income_multiplier
# (constrained to be equal to know_income_multiplier)
'name': 'prior_income_multiplier',
'density': 'uniform',
'lower': known_income_multiplier,
'upper': known_income_multiplier,
'mean': known_income_multiplier,
}
]
# ---------------------------------------------------------------------
# smooth table:
smooth_table = [
{ # smooth_omega_parent
'name': 'smooth_omega_parent',
'age_id': [0],
'time_id': [0],
'fun': fun_omega_parent
},
{ # smooth_income_multiplier
'name': 'smooth_income_multiplier',
'age_id': [0],
'time_id': [0],
'fun': fun_income_multiplier
}
]
# ---------------------------------------------------------------------
# rate table
rate_table = [{
'name': 'omega',
'parent_smooth': 'smooth_omega_parent',
}]
# -------------------------------------------------------------------
# mulcov table: there is one covariate multiplier for income and it
# affects the other cause mortality rate:
mulcov_table = [{
'covariate': 'income',
'type': 'rate_value',
'effected': 'omega',
'group': 'world',
'smooth': 'smooth_income_multiplier'
}]
# -------------------------------------------------------------------
# option_table
option_table = [{
'name': 'parent_node_name',
'value': 'world'
}, {
'name': 'ode_step_size',
'value': '10.0'
}, {
'name': 'age_avg_split',
'value': '5.0'
}, {
'name': 'rate_case',
'value': 'iota_zero_rho_zero'
}, {
'name': 'avgint_extra_columns',
'value': 'avgint_info'
}, {
'name': 'data_extra_columns',
'value': 'c_data_info'
}]
# ---------------------------------------------------------------------
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
file_name = 'get_started.db'
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
def example_db(file_name):
def fun_chi(a, t):
return ('prior_chi', 'prior_dchi', None)
# ----------------------------------------------------------------------
# age table:
age_list = [0.0, 100.0]
#
# time table:
time_list = [1990.0, 2010.0]
#
# integrand table:
integrand_table = [{'name': 'mtexcess'}]
#
# node table:
node_table = [{'name': 'world', 'parent': ''}]
#
# weight table:
weight_table = list()
#
# covariate table: empty
covariate_table = list()
#
# mulcov table: empty
mulcov_table = list()
#
# avgint table: empty
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# data table:
data_table = list()
#
# values that are the same for all data rows
row = {
'meas_value': chi_bar,
'density': 'gaussian',
'weight': '',
'hold_out': False,
'age_lower': age_list[0],
'age_upper': age_list[0],
'time_lower': 2000.,
'time_upper': 2000.,
'integrand': 'mtexcess',
'meas_std': sigma,
'node': 'world'
}
data_table.append(row)
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_chi
'name': 'prior_chi',
'density': 'uniform',
'lower': 0.0,
'upper': 1.0,
'mean': 0.0
},
{ # prior_dchi
'name': 'prior_dchi',
'density': 'gaussian',
'lower': None,
'upper': None,
'mean': 0.0,
'std': delta
}
]
# ----------------------------------------------------------------------
# smooth table
name = 'smooth_chi'
fun = fun_chi
smooth_table = [{
'name': name,
'age_id': [0, 1],
'time_id': [0],
'fun': fun
}]
# ----------------------------------------------------------------------
# rate table:
rate_table = [{
'name': 'chi',
'parent_smooth': 'smooth_chi',
'child_smooth': None
}]
# ----------------------------------------------------------------------
# option_table
option_table = [{
'name': 'rate_case',
'value': 'iota_zero_rho_zero'
}, {
'name': 'parent_node_name',
'value': 'world'
}, {
'name': 'random_seed',
'value': '0'
}, {
'name': 'quasi_fixed',
'value': 'false'
}, {
'name': 'derivative_test_fixed',
'value': 'second-order'
}, {
'name': 'max_num_iter_fixed',
'value': '100'
}, {
'name': 'print_level_fixed',
'value': '0'
}, {
'name': 'tolerance_fixed',
'value': '1e-8'
}]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
# ----------------------------------------------------------------------
return
def example_db(file_name):
def fun_rate_parent(a, t):
return ('prior_rate_parent', None, 'prior_gauss_zero')
# ----------------------------------------------------------------------
# age table
age_list = [0.0, 50.0, 100.0]
#
# time table
time_list = [1995.0, 2005.0, 2015.0]
#
# integrand table
integrand_table = [{
'name': 'prevalence'
}, {
'name': 'Sincidence'
}, {
'name': 'remission'
}, {
'name': 'mtexcess'
}, {
'name': 'mtother'
}]
#
# node table: world -> north_america
# north_america -> (united_states, canada)
node_table = [{
'name': 'world',
'parent': ''
}, {
'name': 'north_america',
'parent': 'world'
}, {
'name': 'united_states',
'parent': 'north_america'
}, {
'name': 'canada',
'parent': 'north_america'
}]
#
# weight table:
weight_table = list()
#
# covariate table: no covriates
covariate_table = list()
#
# mulcov table
mulcov_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# avgint table: same order as list of integrands
avgint_table = list()
# values that are the same for all data rows
row = {
'node': 'canada',
'subgroup': 'world',
'weight': '',
'time_lower': 2000.0,
'time_upper': 2000.0,
'age_lower': 0.0
}
# values that change between rows: (one data point for each integrand)
for avgint_id in range(len(integrand_table)):
integrand = integrand_table[avgint_id]['name']
row['integrand'] = integrand
if integrand == 'prevalence':
# prevalence is measured at age zero
row['age_upper'] = 0.0
else:
# other integrands are averaged from age zero to one hundred
row['age_upper'] = 100.0
avgint_table.append(copy.copy(row))
# ----------------------------------------------------------------------
# data table: same order as list of integrands
data_table = list()
# values that are the same for all data rows
row = {
'node': 'canada',
'subgroup': 'world',
'density': 'gaussian',
'weight': '',
'hold_out': False,
'time_lower': 2000.0,
'time_upper': 2000.0,
'age_lower': 0.0
}
# values that change between rows: (one data point for each integrand)
for integrand_id in range(len(integrand_table)):
rate_id = integrand_id
meas_value = 1e-2 * (rate_id + 1)
meas_std = 0.2 * meas_value
integrand = integrand_table[integrand_id]['name']
row['meas_value'] = meas_value
row['meas_std'] = meas_std
row['integrand'] = integrand
if integrand == 'prevalence':
# prevalence is measured at age zero
row['age_upper'] = 0.0
else:
# other integrands are averaged from age zero to one hundred
row['age_upper'] = 100.0
# data_id = rate_id = integand_id
data_table.append(copy.copy(row))
#
# add one outlyer at end of data table with hold_out true
row['hold_out'] = True # if outlyer were false, fit would fail
row['integrand'] = data_table[0]['integrand']
row['meas_std'] = data_table[0]['meas_std']
row['eta'] = None
row['age_upper'] = data_table[0]['age_upper']
row['meas_value'] = 10. * data_table[0]['meas_value']
data_table.append(copy.copy(row))
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_rate_parent
'name': 'prior_rate_parent',
'density': 'uniform',
'lower': 1e-4,
'upper': 1.0,
'mean': 1e-1,
},{ # prior_gauss_zero
'name': 'prior_gauss_zero',
'density': 'gaussian',
'mean': 0.0,
'std': 1e-2,
}
]
# ----------------------------------------------------------------------
# smooth table
middle_age_id = 1
last_time_id = 2
smooth_table = [{ # smooth_rate_parent
'name': 'smooth_rate_parent',
'age_id': [middle_age_id],
'time_id': [0, last_time_id],
'fun': fun_rate_parent
}]
# ----------------------------------------------------------------------
# rate table
rate_table = [{
'name': 'pini',
'parent_smooth': 'smooth_rate_parent',
}, {
'name': 'iota',
'parent_smooth': 'smooth_rate_parent',
}, {
'name': 'rho',
'parent_smooth': 'smooth_rate_parent',
}, {
'name': 'chi',
'parent_smooth': 'smooth_rate_parent',
}, {
'name': 'omega',
'parent_smooth': 'smooth_rate_parent',
}]
# ----------------------------------------------------------------------
# option_table
option_table = [{
'name': 'parent_node_name',
'value': 'canada'
}, {
'name': 'ode_step_size',
'value': '10.0'
}, {
'name': 'random_seed',
'value': '0'
}, {
'name': 'rate_case',
'value': 'iota_pos_rho_pos'
}, {
'name': 'quasi_fixed',
'value': 'true'
}, {
'name': 'derivative_test_fixed',
'value': 'first-order'
}, {
'name': 'max_num_iter_fixed',
'value': '100'
}, {
'name': 'print_level_fixed',
'value': '0'
}, {
'name': 'tolerance_fixed',
'value': '1e-10'
}, {
'name': 'derivative_test_random',
'value': 'second-order'
}, {
'name': 'max_num_iter_random',
'value': '100'
}, {
'name': 'print_level_random',
'value': '0'
}, {
'name': 'tolerance_random',
'value': '1e-10'
}]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
# ----------------------------------------------------------------------
n_smooth = len(smooth_table)
rate_true = []
for rate_id in range(len(data_table)):
# for this particular example
data_id = rate_id
meas_value = data_table[data_id]['meas_value']
rate_true.append(meas_value)
#
return (n_smooth, rate_true)
def example_db (file_name) :
# ----------------------------------------------------------------------
# time table
time_list = [ 1990.0, 2015.0 ]
#
# integrand table
integrand_table = [
{ 'name':'Sincidence', 'minimum_meas_cv':0.0 }
]
#
# node table: world
node_table = [
{ 'name':'world', 'parent':'' }
]
#
# weight table: The constant function 1.0 (one age and one time point)
weight_table = [
{ 'name':'weight_one', 'age_id':[1], 'time_id':[1], 'fun':fun_one }
]
#
# covariate table:
covariate_table = list()
#
# mulcov table
mulcov_table = list()
#
# ----------------------------------------------------------------------
# data table: order for value_list
# 'integrand', 'density', 'node', 'weight', 'hold_out',
# 'meas_value', 'meas_std', 'age_lower', 'age_upper',
# 'time_lower', 'time_upper', 'sex', 'one', 'subgroup'
data_table = list()
#
# Sincidence
value_list = [
'Sincidence', 'log_gaussian', 'world', 'weight_one', 0,
4.0e-3, 2.0e-3, 1e-6, 40, 40,
1995, 1995, 0.0, 1, 'world'
]
data_table.append( data_list2dict( value_list ) );
#
value_list = [
'Sincidence', 'log_gaussian', 'world', 'weight_one', 0,
4.0e-2, 2.0e-2, 1e-6, 100, 100,
1995, 1995, 0.0, 1, 'world'
]
data_table.append( data_list2dict( value_list ) );
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_zero
'name': 'prior_zero',
'density': 'uniform',
'lower': 0.0,
'upper': 0.0,
'mean': 0.0,
'std': None,
'eta': None
},{ # prior_rate
'name': 'prior_rate',
'density': 'log_gaussian',
'lower': 1e-5,
'upper': 1e+0,
'mean': 1e-2,
'std': 1e-0,
'eta': 1e-6
},{ # prior_difference
'name': 'prior_difference',
'density': 'log_gaussian',
'lower': None,
'upper': None,
'mean': 0.0,
'std': 1.0,
'eta': 1e-6
}
]
# ----------------------------------------------------------------------
# smooth table
middle_age_id = 1
last_time_id = 2
smooth_table = [
{
# smooth_rate
'name': 'smooth_rate',
'age_id': range(len(age_list)),
'time_id': [0],
'mulstd_value_prior_name': None,
'mulstd_dage_prior_name': None,
'mulstd_dtime_prior_name': None,
'fun': fun_rate
}
]
# ----------------------------------------------------------------------
# rate table
rate_table = [
{
'name': 'pini',
'parent_smooth': None,
'child_smooth': None,
'child_nslist': None
},{
'name': 'iota',
'parent_smooth': 'smooth_rate',
'child_smooth': None,
'child_nslist': None
},{
'name': 'rho',
'parent_smooth': None,
'child_smooth': None,
'child_nslist': None
},{
'name': 'chi',
'parent_smooth': None,
'child_smooth': None,
'child_nslist': None
},{
'name': 'omega',
'parent_smooth': None,
'child_smooth': None,
'child_nslist': None
}
]
# ----------------------------------------------------------------------
# option_table
option_table = [
{ 'name':'parent_node_name', 'value':'world' },
{ 'name':'rate_case', 'value':'iota_pos_rho_zero' },
#
{ 'name':'quasi_fixed', 'value':'false' },
{ 'name':'derivative_test_fixed', 'value':'first-order' },
#
{ 'name':'derivative_test_random', 'value':'second-order' }
]
# ----------------------------------------------------------------------
# avgint table: same order as list of integrands
avgint_table = list()
# ----------------------------------------------------------------------
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [ { 'subgroup':'world', 'group':'world' } ]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(
file_name,
age_list,
time_list,
integrand_table,
node_table,
subgroup_table,
weight_table,
covariate_table,
avgint_table,
data_table,
prior_table,
smooth_table,
nslist_table,
rate_table,
mulcov_table,
option_table
)
return
def example_db(file_name):
def fun_prevalence(a, t):
return ('prior_prevalence', 'prior_dage', 'prior_dtime')
def fun_mulcov(a, t):
return ('prior_mulcov', None, None)
# ----------------------------------------------------------------------
# age table:
age_list = [0.0, 5.0, 10.0, 20.0, 40.0, 60.0, 80.0, 100.0]
#
# time table:
time_list = [1990.0, 2200.0]
#
# integrand table:
integrand_table = [{'name': 'remission'}]
#
# node table:
node_table = [{'name': 'world', 'parent': ''}]
#
# weight table:
weight_table = list()
#
# covariate table:
covariate_table = [{'name': 'income', 'reference': income_reference}]
#
# mulcov table:
if income_mulcov_type == 'rate_value':
effected = 'rho'
else:
assert income_mulcov_type == 'meas_value'
effected = 'remission'
mulcov_table = [{
'covariate': 'income',
'type': income_mulcov_type,
'effected': effected,
'group': 'world',
'smooth': 'smooth_mulcov'
}]
#
# avgint table: empty
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# data table:
data_table = list()
#
# values that are the same for all data rows
row = {
'node': 'world',
'subgroup': 'world',
'integrand': 'remission',
'density': 'log_gaussian',
'weight': '',
'hold_out': False,
'time_lower': 2000.,
'time_upper': 2000.,
'eta': prevalence_eta
}
n_data = 2000
n_income = 30
for data_id in range(n_data):
age = 100.0 * data_id / float(n_data - 1)
income = (data_id % n_income) / float(n_income - 1)
income_effect = (income - income_reference) * income_multiplier
eta = prevalence_eta
sigma = prevalence_sigma
log_noise = random.gauss(0, sigma)
y = exp(log_noise) * Prevalence(age) * exp(income_effect)
delta = (exp(sigma) - 1) * (y + eta)
row['age_lower'] = age
row['age_upper'] = age
row['income'] = income
row['meas_value'] = y
row['meas_std'] = delta
data_table.append(copy.copy(row))
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_prevalence
'name': 'prior_prevalence',
'density': 'uniform',
'lower': 0.0,
'upper': 1.0,
'mean': 0.01,
},{ # prior_dage
'name': 'prior_dage',
'density': 'log_gaussian',
'mean': 0.0,
'std': 2.0,
'eta': prevalence_eta,
},{ # prior_dtime
'name': 'prior_dtime',
'density': 'log_gaussian',
'mean': 0.0,
'std': 0.1,
'eta': prevalence_eta,
}, { # prior_mulcov
'name': 'prior_mulcov',
'density': 'uniform',
'lower': -2.0,
'upper': +2.0,
'mean': 0.0,
}
]
# ----------------------------------------------------------------------
# smooth table
smooth_table = list()
#
name = 'smooth_prevalence'
fun = fun_prevalence
age_id = range(len(age_list))
time_id = range(len(time_list))
smooth_table.append({
'name': name,
'age_id': age_id,
'time_id': time_id,
'fun': fun
})
name = 'smooth_mulcov'
fun = fun_mulcov
smooth_table.append({
'name': name,
'age_id': [0],
'time_id': [0],
'fun': fun
})
# ----------------------------------------------------------------------
# rate table:
rate_table = [{
'name': 'rho',
'parent_smooth': 'smooth_prevalence',
'child_smooth': None,
}]
# ----------------------------------------------------------------------
# option_table
option_table = [{
'name': 'rate_case',
'value': 'no_ode'
}, {
'name': 'parent_node_name',
'value': 'world'
}, {
'name': 'ode_step_size',
'value': '5.0'
}, {
'name': 'random_seed',
'value': random_seed
}, {
'name': 'quasi_fixed',
'value': 'false'
}, {
'name': 'derivative_test_fixed',
'value': 'first-order'
}, {
'name': 'max_num_iter_fixed',
'value': '100'
}, {
'name': 'print_level_fixed',
'value': '0'
}, {
'name': 'tolerance_fixed',
'value': '1e-8'
}, {
'name': 'derivative_test_random',
'value': 'second-order'
}, {
'name': 'max_num_iter_random',
'value': '100'
}, {
'name': 'print_level_random',
'value': '0'
}, {
'name': 'tolerance_random',
'value': '1e-8'
}]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
# ----------------------------------------------------------------------
return
def example_db(file_name):
# note that the a, t values are not used for this case
def fun_rate_child(a, t):
return ('prior_gauss_zero', 'prior_gauss_zero', 'prior_gauss_zero')
def fun_rate_parent(a, t):
return ('prior_value_parent', 'prior_diff_parent', 'prior_diff_parent')
# ----------------------------------------------------------------------
# age table
age_list = [0.0, 50.0, 100.0]
#
# time table
time_list = [1995.0, 2005.0, 2015.0]
#
# integrand table
integrand_table = [{
'name': 'Sincidence'
}, {
'name': 'remission'
}, {
'name': 'mtexcess'
}, {
'name': 'mtother'
}]
#
# node table: world -> north_america
# north_america -> (united_states, canada)
node_table = [{
'name': 'world',
'parent': ''
}, {
'name': 'north_america',
'parent': 'world'
}, {
'name': 'united_states',
'parent': 'north_america'
}, {
'name': 'canada',
'parent': 'north_america'
}]
#
# weight table:
weight_table = list()
#
# covariate table: no covriates
covariate_table = list()
#
# mulcov table
mulcov_table = list()
#
# avgint table: empty
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# data table: same order as list of integrands
data_table = list()
# values that are the same for all data rows
row = {
'node': 'world',
'subgroup': 'world',
'density': 'gaussian',
'weight': '',
'hold_out': False,
'time_lower': 1995.0,
'time_upper': 1995.0,
'age_lower': 0.0,
'age_upper': 0.0
}
# values that change between rows: (one data point for each integrand)
for integrand_id in range(len(integrand_table)):
rate_id = integrand_id
meas_value = 0.05
meas_std = 0.2 * meas_value
integrand = integrand_table[integrand_id]['name']
row['meas_value'] = meas_value
row['meas_std'] = meas_std
row['integrand'] = integrand
# data_id = rate_id = integand_id
data_table.append(copy.copy(row))
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_gauss_zero
'name': 'prior_gauss_zero',
'density': 'gaussian',
'mean': 0.0,
'std': 0.01,
},{ # prior_value_parent
'name': 'prior_value_parent',
'density': 'uniform',
'lower': 0.01,
'upper': 1.00,
'mean': 0.1,
},{ # prior_diff_parent
'name': 'prior_diff_parent',
'density': 'gaussian',
'lower': 0.01,
'upper': 1.0,
'mean': 0.01,
'std': 0.01,
}
]
# ----------------------------------------------------------------------
# smooth table
middle_age_id = 1
middle_time_id = 1
last_age_id = 2
last_time_id = 2
smooth_table = [
{ # smooth_rate_child
'name': 'smooth_rate_child',
'age_id': [0, last_age_id],
'time_id': [0, last_time_id],
'fun': fun_rate_child
},
{ # smooth_rate_parent
'name': 'smooth_rate_parent',
'age_id': [0, last_age_id],
'time_id': [0, last_time_id],
'fun': fun_rate_parent
}
]
# ----------------------------------------------------------------------
# rate table
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_rate_parent',
'child_smooth': 'smooth_rate_child',
}, {
'name': 'rho',
'parent_smooth': 'smooth_rate_parent',
'child_smooth': 'smooth_rate_child',
}, {
'name': 'chi',
'parent_smooth': 'smooth_rate_parent',
'child_smooth': 'smooth_rate_child',
}, {
'name': 'omega',
'parent_smooth': 'smooth_rate_parent',
'child_smooth': 'smooth_rate_child',
}]
# ----------------------------------------------------------------------
# option_table
option_table = [{
'name': 'parent_node_name',
'value': 'world'
}, {
'name': 'ode_step_size',
'value': '10.0'
}, {
'name': 'random_seed',
'value': '0'
}, {
'name': 'rate_case',
'value': 'iota_pos_rho_pos'
}, {
'name': 'quasi_fixed',
'value': 'true'
}, {
'name': 'derivative_test_fixed',
'value': 'first-order'
}, {
'name': 'max_num_iter_fixed',
'value': '100'
}, {
'name': 'print_level_fixed',
'value': '0'
}, {
'name': 'tolerance_fixed',
'value': '1e-10'
}, {
'name': 'derivative_test_random',
'value': 'second-order'
}, {
'name': 'max_num_iter_random',
'value': '100'
}, {
'name': 'print_level_random',
'value': '0'
}, {
'name': 'tolerance_random',
'value': '1e-10'
}]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
def initDatabase(self, max_iter=500):
if len(self.age_list) == 0:
max_age = -float('inf')
min_age = float('inf')
for i in range(self.n):
max_age = max(max_age, self.data.loc[i, 'age_end'])
min_age = min(min_age, self.data.loc[i, 'age_start'])
age_list = [int(round(x)) for x in np.linspace(min_age,max_age,\
round((max_age-min_age)/5)+1)]
age_list = sorted(list(set(age_list)))
#if len(age_list) == 1:
# age_list.insert(0,age_list[0]-1)
self.age_list = age_list
if len(self.time_list) == 0:
max_time = -float('inf')
min_time = float('inf')
for i in range(self.n):
max_time = max(max_time, self.data.loc[i, 'year_end'])
min_time = min(min_time, self.data.loc[i, 'year_start'])
time_list = [int(round(x)) for x in np.linspace(min_time, max_time,\
round((max_time - min_time)/3+1))]
time_list = sorted(list(set(time_list)))
self.time_list = time_list
#print(self.age_list)
#print(self.time_list)
avgint_table = list()
nslist_table = dict() # smoothing
integrand_table = []
for intg in self.integrand:
integrand_table.append({'name': intg})
node_table = [{'name': 'world', 'parent': ''}]
weight_table = [{
'name': 'constant',
'age_id': range(len(self.age_list)),
'time_id': range(len(self.time_list)),
'fun': lambda a, t: 1.0
}]
rate_table = list()
for rate in self.rates:
rate_table.append({
'name': rate,
'parent_smooth': 'smooth_rate_' + rate
})
covariate_table = list()
for cov in self.covariates:
covariate_table.append({'name': cov['name'], 'reference': 0.0})
mulcov_table = list()
for cov in self.covariates:
mulcov_table.append({
'covariate': cov['name'],
'type': cov['type'],
'effected': cov['effected'],
'smooth': 'smooth_mulcov_' + cov['name']
})
smooth_table = list()
for rate in self.rates:
smooth_table.append({
'name':
'smooth_rate_' + rate,
'age_id':
range(len(self.age_list)),
'time_id':
range(len(self.time_list)),
'fun':
lambda a, t, r=rate:
('value_prior_' + r, 'dage_prior_' + r, 'dtime_prior_' + r)
})
for cov in self.covariates:
name = cov['name']
smooth_table.append({
'name':
'smooth_mulcov_' + cov['name'],
'age_id':
range(len(self.age_list)),
'time_id':
range(len(self.time_list)),
'fun':
lambda a, t, name=name: ('value_prior_' + name, 'dage_prior_' +
name, 'dtime_prior_' + name)
})
#for row in smooth_table:
# print(row['fun'](0,0))
prior_table = []
for i in range(len(self.rates)):
prior_table.append({'name': 'value_prior_' + self.rates[i]})
prior_table[-1].update(self.rate_priors[i][0])
prior_table.append({'name': 'dage_prior_' + self.rates[i]})
prior_table[-1].update(self.rate_priors[i][1])
prior_table.append({'name': 'dtime_prior_' + self.rates[i]})
prior_table[-1].update(self.rate_priors[i][2])
for i in range(len(self.covariates)):
prior_table.append(
{'name': 'value_prior_' + self.covariates[i]['name']})
prior_table[-1].update(self.cov_priors[i][0])
prior_table.append(
{'name': 'dage_prior_' + self.covariates[i]['name']})
prior_table[-1].update(self.cov_priors[i][1])
prior_table.append(
{'name': 'dtime_prior_' + self.covariates[i]['name']})
prior_table[-1].update(self.cov_priors[i][2])
data_table = list()
row = {
'node': 'world',
'weight': 'constant',
'hold_out': False,
}
row.update(self.meas_noise_density)
for data_id in range(self.n):
if self.data.loc[data_id, 'measure'] in self.integrand:
row['integrand'] = self.data.loc[data_id, 'measure']
for k, v in self.meas_noise_density[row['integrand']].items():
row[k] = v
row['meas_value'] = self.data.loc[data_id, 'meas_value']
row['meas_std'] = self.data.loc[data_id, 'meas_std']
row['age_lower'] = self.data.loc[data_id, 'age_start']
row['age_upper'] = self.data.loc[data_id, 'age_end']
row['time_lower'] = self.data.loc[data_id, 'year_start']
row['time_upper'] = self.data.loc[data_id, 'year_end']
for cov in self.covariates:
row[cov['name']] = self.data.loc[data_id, cov['name']]
data_table.append(copy.copy(row))
option_table = [
{
'name': 'parent_node_name',
'value': 'world'
},
{
'name': 'ode_step_size',
'value': '10.0'
},
{
'name': 'quasi_fixed',
'value': 'false'
},
{
'name': 'max_num_iter_fixed',
'value': max_iter
},
{
'name': 'print_level_fixed',
'value': '5'
},
{
'name': 'tolerance_fixed',
'value': '1e-8'
},
]
if self.integrand == ['Sincidence']:
option_table.append({
'name': 'rate_case',
'value': 'iota_pos_rho_zero'
})
elif self.integrand == 'remission':
option_table.append({
'name': 'rate_case',
'value': 'iota_zero_rho_pos'
})
else:
option_table.append({
'name': 'rate_case',
'value': 'iota_pos_rho_pos'
})
option_name_id = {}
for i in range(len(option_table)):
option_name_id[option_table[i]['name']] = i
for option in self.options:
if option['name'] in option_name_id:
option_table[option_name_id[
option['name']]]['value'] = option['value']
else:
option_table.append(option)
dismod_at.create_database(self.path, self.age_list, self.time_list,
integrand_table, node_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
command = [program, self.path, 'init']
print(' '.join(command))
flag = subprocess.call(command)
if flag != 0:
sys.exit('The dismod_at init command failed')
dismod_at.db2csv_command(self.path)
def example_db(file_name):
# note that the a, t values are not used for this case
def fun_iota(a, t):
return ('prior_iota', None, None)
def fun_gamma(a, t):
return ('prior_gamma', None, None)
# ----------------------------------------------------------------------
# age table:
age_list = [0.0, 100.0]
#
# time table:
time_list = [1990.0, 2010.0]
#
# integrand table:
integrand_table = [{'name': 'Sincidence'}]
#
# node table:
node_table = [{'name': 'world', 'parent': ''}]
#
# weight table:
weight_table = list()
#
# covariate table:
covariate_table = [{'name': 'one', 'reference': 0.0}]
#
# mulcov table:
mulcov_table = [{
'covariate': 'one',
'type': 'meas_noise',
'effected': 'Sincidence',
'group': 'world',
'smooth': 'smooth_gamma'
}]
#
# avgint table: empty
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# data table:
data_table = list()
# values that are the same for all data rows
row = {
'weight': '',
'hold_out': False,
'node': 'world',
'subgroup': 'world',
'one': 1.0,
'age_lower': 50.0,
'age_upper': 50.0,
'time_lower': 2000.,
'time_upper': 2000.,
'integrand': 'Sincidence',
'density': data_density,
'meas_std': meas_std,
'eta': iota_true / 100.0,
'nu': 10
}
# values that change between rows:
for data_id in range(n_data):
if data_id % 2 == 0:
row['meas_value'] = 0.9 * iota_true
else:
row['meas_value'] = 1.1 * iota_true
data_table.append(copy.copy(row))
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_iota
'name': 'prior_iota',
'density': 'uniform',
'lower': iota_true / 100.0,
'upper': 1.0,
'mean': iota_true / 10.0
},
{ # prior_gamma
'name': 'prior_gamma',
'density': 'uniform',
'lower': 0.0,
'upper': 10.0 * gamma_true(),
'mean': gamma_true() / 10.0
}
]
# ----------------------------------------------------------------------
# smooth table
name = 'smooth_iota'
fun = fun_iota
age_id = 0
time_id = 0
smooth_table = [{
'name': name,
'age_id': [age_id],
'time_id': [time_id],
'fun': fun
}]
name = 'smooth_iota'
#
name = 'smooth_gamma'
fun = fun_gamma
smooth_table.append({
'name': name,
'age_id': [age_id],
'time_id': [time_id],
'fun': fun
})
# ----------------------------------------------------------------------
# rate table:
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_iota',
'child_smooth': None
}]
# ----------------------------------------------------------------------
# option_table
option_table = [
{
'name': 'meas_noise_effect',
'value': meas_noise_effect
},
{
'name': 'rate_case',
'value': 'iota_pos_rho_zero'
},
{
'name': 'parent_node_name',
'value': 'world'
},
{
'name': 'random_seed',
'value': '0'
},
]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
# ----------------------------------------------------------------------
return
def example_db (file_name) :
# note that the a, t values are not used for this case
def fun_rate_child(a, t) :
return ('prior_iota_child', None, None)
def fun_iota_parent(a, t) :
return ('prior_iota_parent', None, None)
def fun_mulcov(a, t) :
return ('prior_mulcov', None, None)
# ----------------------------------------------------------------------
# age table:
age_list = [ 0.0, 5.0, 15.0, 35.0, 50.0, 75.0, 90.0, 100.0 ]
#
# time table:
time_list = [ 1990.0, 2000.0, 2010.0, 2200.0 ]
#
# integrand table:
integrand_table = [
{ 'name':'prevalence' }
]
#
# node table:
node_table = [ { 'name':'world', 'parent':'' } ]
for i in range(n_children) :
name = 'child_' + str(i + 1)
node_table.append( { 'name':name, 'parent':'world' } )
#
# weight table:
weight_table = list()
#
# covariate table:
covariate_table = [
{'name':'income', 'reference':0.5}
]
#
# mulcov table:
mulcov_table = [
{
'covariate': 'income',
'type': 'rate_value',
'effected': 'iota',
'group': 'world',
'smooth': 'smooth_mulcov'
}
]
#
# avgint table: empty
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# data table:
data_table = list()
# values that are the same for all data rows
row = {
'meas_value': 0.0, # not used (will be simulated)
'density': 'gaussian',
'weight': '',
'hold_out': False,
'time_lower': 2000.,
'time_upper': 2000.
}
# values that change between rows:
for data_id in range( n_data ) :
fraction = data_id / float(n_data-1)
age = age_list[0] + (age_list[-1] - age_list[0])*fraction
row['age_lower'] = age
row['age_upper'] = age
row['node'] = 'child_' + str( (data_id % n_children) + 1 )
row['income'] = fraction
row['integrand'] = integrand_table[0]['name']
row['meas_std'] = 1e-3
data_table.append( copy.copy(row) )
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_iota_child
'name': 'prior_iota_child',
'density': 'gaussian',
'mean': 0.0,
'std': 0.5,
},{ # prior_iota_parent
'name': 'prior_iota_parent',
'density': 'uniform',
'lower': iota_parent_true / 100.,
'upper': 1.0,
'mean': 0.1,
},{ # prior_mulcov
'name': 'prior_mulcov',
'density': 'uniform',
'lower': -2.0,
'upper': +2.0,
'mean': 0.0,
}
]
# ----------------------------------------------------------------------
# smooth table
name = 'smooth_rate_child'
fun = fun_rate_child
age_id = int( len( age_list ) / 2 )
time_id = int( len( time_list ) / 2 )
smooth_table = [
{'name':name, 'age_id':[age_id], 'time_id':[time_id], 'fun':fun }
]
name = 'smooth_iota_parent'
fun = fun_iota_parent
smooth_table.append(
{'name':name, 'age_id':[age_id], 'time_id':[time_id], 'fun':fun }
)
name = 'smooth_mulcov'
fun = fun_mulcov
smooth_table.append(
{'name':name, 'age_id':[age_id], 'time_id':[time_id], 'fun':fun }
)
# no standard deviation multipliers
for dictionary in smooth_table :
for name in [ 'value' , 'dage', 'dtime' ] :
key = 'mulstd_' + name + '_prior_name'
value = None
dictionary[key] = value
# ----------------------------------------------------------------------
# rate table:
rate_table = [
{ 'name': 'iota',
'parent_smooth': 'smooth_iota_parent',
'child_smooth': 'smooth_rate_child',
}
]
# ----------------------------------------------------------------------
# option_table
option_table = [
{ 'name':'rate_case', 'value':'iota_pos_rho_zero' },
{ 'name':'parent_node_name', 'value':'world' },
{ 'name':'ode_step_size', 'value':'10.0' },
{ 'name':'random_seed', 'value':'0' },
{ 'name':'zero_sum_child_rate', 'value':'iota' },
{ 'name':'quasi_fixed', 'value':'true' },
{ 'name':'derivative_test_fixed', 'value':'first-order' },
{ 'name':'max_num_iter_fixed', 'value':'100' },
{ 'name':'print_level_fixed', 'value':'0' },
{ 'name':'tolerance_fixed', 'value':'1e-8' },
{ 'name':'derivative_test_random', 'value':'second-order' },
{ 'name':'max_num_iter_random', 'value':'100' },
{ 'name':'print_level_random', 'value':'0' },
{ 'name':'tolerance_random', 'value':'1e-8' }
]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [ { 'subgroup':'world', 'group':'world' } ]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(
file_name,
age_list,
time_list,
integrand_table,
node_table,
subgroup_table,
weight_table,
covariate_table,
avgint_table,
data_table,
prior_table,
smooth_table,
nslist_table,
rate_table,
mulcov_table,
option_table
)
# ----------------------------------------------------------------------
return
def example_db(file_name):
def fun_income(a, t):
return ('prior_income_value', 'prior_income_dage',
'prior_income_dtime')
def fun_iota_parent(a, t):
return ('prior_iota_parent', None, None)
# ----------------------------------------------------------------------
# age table
#
# time table
node_table = [
{
'name': 'north_america',
'parent': ''
},
]
#
# weight table:
weight_table = list()
# integrand table
integrand_table = [{'name': 'Sincidence'}, {'name': 'mulcov_0'}]
#
# covariate table:
covariate_table = [{'name': 'income', 'reference': income_reference}]
#
# mulcov table: one multiplier with mulcov_id = 0
mulcov_table = [{
'covariate': 'income',
'type': 'rate_value',
'effected': 'iota',
'group': 'world',
'smooth': 'smooth_income'
}]
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_iota_parent
'name': 'prior_iota_parent',
'density': 'uniform',
'lower': iota_reference,
'upper': iota_reference,
'mean': iota_reference,
},{ # prior_income_value
'name': 'prior_income_value',
'density': 'uniform',
'mean': 0.0,
'lower': -1.0,
'upper': +1.0,
},{ # prior_income_dage
'name': 'prior_income_dage',
'density': 'uniform',
'mean': 0.0,
},{ # prior_income_dtime
'name': 'prior_income_dtime',
'density': 'uniform',
'mean': 0.0,
}
]
# ----------------------------------------------------------------------
# smooth table
smooth_table = [
{ # smooth_iota_parent
'name': 'smooth_iota_parent',
'age_id': [0],
'time_id': [0],
'fun': fun_iota_parent
},
{ # smooth_income
'name': 'smooth_income',
'age_id': [0, 1],
'time_id': [0, 1],
'fun': fun_income
}
]
# ----------------------------------------------------------------------
# rate table
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_iota_parent',
}]
# --------------------------------------------------------------------
# option_table
option_table = [
{
'name': 'parent_node_name',
'value': 'north_america'
},
{
'name': 'rate_case',
'value': 'iota_pos_rho_zero'
},
{
'name': 'quasi_fixed',
'value': 'true'
},
{
'name': 'max_num_iter_fixed',
'value': '30'
},
{
'name': 'print_level_fixed',
'value': '0'
},
{
'name': 'tolerance_fixed',
'value': '1e-10'
},
]
# ----------------------------------------------------------------------
# data table:
data_table = list()
row = {
'node': 'north_america',
'subgroup': 'world',
'density': 'gaussian',
'weight': '',
'hold_out': False,
'integrand': 'Sincidence',
'income': 2.0,
}
for age in age_list:
for time in time_list:
alpha = alpha_true(age, time)
effect = alpha * (row['income'] - income_reference)
iota = iota_reference * math.exp(effect)
row['age_lower'] = age
row['age_upper'] = age
row['time_lower'] = time
row['time_upper'] = time
row['meas_value'] = iota
row['meas_std'] = iota / 10.0
data_table.append(copy.copy(row))
# ----------------------------------------------------------------------
# avgint table:
avgint_table = list()
# values that are the same for all rows
does_not_matter = 5.0
row = {
'node': 'north_america',
'subgroup': 'world',
'integrand': 'mulcov_0',
'weight': '',
'income': does_not_matter,
}
for age in age_list:
for time in time_list:
row['age_lower'] = age
row['age_upper'] = age
row['time_lower'] = time
row['time_upper'] = time
avgint_table.append(copy.copy(row))
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
def example_db (file_name) :
# note that the a, t values are not used for this case
def fun_iota(a, t) :
if (a, t) == (0, 2000) :
return ('prior_value_0_2000', 'prior_diff', 'prior_diff')
elif (a, t) == (100, 2000) :
return ('prior_value_100_2000', 'prior_diff', 'prior_diff')
elif (a, t) == (0, 2020) :
return ('prior_value_0_2020', 'prior_diff', 'prior_diff')
elif (a, t) == (100, 2020) :
return ('prior_value_100_2020', 'prior_diff', 'prior_diff')
else :
assert False
# ----------------------------------------------------------------------
# age table:
age_list = [ 0.0, 100.0 ]
#
# time table:
time_list = [ 2000.0, 2020.0 ]
#
# integrand table:
integrand_table = [
{ 'name': 'Sincidence' },
{ 'name': 'prevalence' },
]
#
# node table:
node_table = [ { 'name':'world', 'parent':'' } ]
#
# weight table:
weight_table = list()
#
# covariate table:
covariate_table = list()
#
# mulcov table:
mulcov_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# data table:
data_table = list()
# values that are the same for all data rows
age_lower = 10.0
age_upper = 90.0
time_lower = 2000.0
time_upper = 2020.0
row = {
'weight': '',
'hold_out': False,
'node': 'world',
'subgroup': 'world',
'density': 'gaussian',
'meas_std': 0.01,
'age_lower': age_lower,
'age_upper' : age_upper,
'time_lower' : time_lower,
'time_upper' : time_upper,
}
#
max_step = 1.0
age_grid = uniform_grid(age_lower, age_upper, max_step)
time_grid = uniform_grid(time_lower, time_upper, max_step)
grid = { 'age' : age_grid, 'time' : time_grid }
#
# Sincidence
meas_value = average_integrand('Sincidence', grid)
row['integrand'] = 'Sincidence'
row['meas_value'] = meas_value
data_table.append( copy.copy(row) )
#
# prevalence
meas_value = average_integrand('prevalence', grid)
row['integrand'] = 'prevalence'
row['meas_value'] = meas_value
data_table.append( copy.copy(row) )
#
# avgint table:
avgint_table = data_table
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_value_0_2000
'name': 'prior_value_0_2000',
'density': 'uniform',
'lower': iota_true(0, 2000) / 100.0,
'upper': iota_true(0, 2000) * 100.0,
'mean': iota_true(0, 2000)
},{ # prior_value_100_2000
'name': 'prior_value_100_2000',
'density': 'uniform',
'lower': iota_true(100, 2000) / 100.0,
'upper': iota_true(100, 2000) * 100.0,
'mean': iota_true(100, 2000)
},{ # prior_value_0_2020
'name': 'prior_value_0_2020',
'density': 'uniform',
'lower': iota_true(0, 2020) / 100.0,
'upper': iota_true(0, 2020) * 100.0,
'mean': iota_true(0, 2020)
},{ # prior_value_100_2020
'name': 'prior_value_100_2020',
'density': 'uniform',
'lower': iota_true(100, 2020) / 100.0,
'upper': iota_true(100, 2020) * 100.0,
'mean': iota_true(100, 2020)
},{ # prior_diff
'name': 'prior_diff',
'density': 'uniform',
'mean': 0.0
}
]
# ----------------------------------------------------------------------
# smooth table
name = 'smooth_iota'
fun = fun_iota
age_id = [0, 1]
time_id = [0, 1]
smooth_table = [
{'name':name, 'age_id':age_id, 'time_id':time_id, 'fun':fun }
]
name = 'smooth_iota'
# ----------------------------------------------------------------------
# rate table:
rate_table = [
{ 'name': 'iota',
'parent_smooth': 'smooth_iota',
'child_smooth': None
}
]
# ----------------------------------------------------------------------
# option_table
option_table = [
{ 'name':'rate_case', 'value':'iota_pos_rho_zero' },
{ 'name':'parent_node_name', 'value':'world' },
{ 'name':'ode_step_size', 'value':'1.0' },
]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [ { 'subgroup':'world', 'group':'world' } ]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(
file_name,
age_list,
time_list,
integrand_table,
node_table,
subgroup_table,
weight_table,
covariate_table,
avgint_table,
data_table,
prior_table,
smooth_table,
nslist_table,
rate_table,
mulcov_table,
option_table
)
# ----------------------------------------------------------------------
return
def example_db(file_name):
def fun_rate_world(a, t):
return ('prior_rate_world', None, 'prior_world_diff')
import dismod_at
# ----------------------------------------------------------------------
# age table
age_list = [0.0, 100.0]
#
# time table
time_list = [1995.0, 2015.0]
#
# integrand table
integrand_table = list()
#
# node table:
node_table = [{'name': 'world', 'parent': ''}]
#
# weight table
weight_table = list()
#
# covariate table: no covriates
covariate_table = list()
#
# mulcov table
mulcov_table = list()
#
# avgint table: same order as list of integrands
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
#
# data table:
data_table = list()
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_rate_world
'name': 'prior_rate_world',
'density': 'gaussian',
'mean': omega_world_mean,
'std': omega_world_mean
},
{ # prior_world_diff
'name': 'prior_world_diff',
'density': 'gaussian',
'mean': 0.0,
'std': omega_world_mean
}
]
# ----------------------------------------------------------------------
# smooth table
last_time_id = len(time_list) - 1
smooth_table = [{ # smooth_rate_world
'name': 'smooth_rate_world',
'age_id': [0],
'time_id': [0, last_time_id],
'fun': fun_rate_world
}]
# ----------------------------------------------------------------------
# rate table
rate_table = [{
'name': 'omega',
'parent_smooth': 'smooth_rate_world',
'child_smooth': None,
}]
# ----------------------------------------------------------------------
# option_table
option_table = [{
'name': 'parent_node_name',
'value': 'world'
}, {
'name': 'rate_case',
'value': 'iota_zero_rho_zero'
}, {
'name': 'quasi_fixed',
'value': 'true'
}, {
'name': 'derivative_test_fixed',
'value': 'first-order'
}, {
'name': 'max_num_iter_fixed',
'value': '100'
}, {
'name': 'print_level_fixed',
'value': '0'
}, {
'name': 'tolerance_fixed',
'value': '1e-11'
}, {
'name': 'derivative_test_random',
'value': 'second-order'
}, {
'name': 'max_num_iter_random',
'value': '100'
}, {
'name': 'print_level_random',
'value': '0'
}, {
'name': 'tolerance_random',
'value': '1e-11'
}]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
# ----------------------------------------------------------------------
return
def example_db (file_name) :
def fun_rate_parent(a, t) :
return ('prior_rate_parent', 'prior_none', 'prior_none')
# ----------------------------------------------------------------------
# age table
age_list = list( range(0, 101, 10) )
age_index = list( range(1, len(age_list) - 1, 1) )
#
# time table
time_list = list( range(1980, 2021, 10) )
time_index = list( range(1, len(time_list) - 1, 1) )
#
# integrand table
integrand_table = [
{ 'name':'Sincidence' },
{ 'name':'remission' },
{ 'name':'mtexcess' },
]
#
# node table: world -> north_america
# north_america -> (united_states, canada)
node_table = [
{ 'name':'world', 'parent':'' },
{ 'name':'north_america', 'parent':'world' },
{ 'name':'united_states', 'parent':'north_america' },
{ 'name':'canada', 'parent':'north_america' }
]
#
# weight table:
weight_table = list()
#
# covariate table: no covriates
covariate_table = list()
#
# mulcov table
mulcov_table = list()
#
# nslist_table:
nslist_table = dict()
#
# avgint table:
avgint_table = list()
# ----------------------------------------------------------------------
# data table: same order as list of integrands
data_table = list()
# values that are the same for all data rows
row = {
'node': 'canada',
'subgroup': 'world',
'density': 'gaussian',
'weight': '',
'hold_out': False,
}
# values that change between rows: (one data point for each integrand)
for integrand_id in range( len(integrand_table) ) :
integrand_name = integrand_table[integrand_id]['name']
rate_name = integrand2rate[integrand_name]
#
for age_id in age_index :
for time_id in time_index :
age = age_list[age_id]
time = time_list[time_id]
true_value = rate_true(rate_name, age, time)
meas_std = meas_cv * true_value
row['age_lower'] = age
row['age_upper'] = age
row['time_lower'] = time
row['time_upper'] = time
row['integrand'] = integrand_name
row['meas_value'] = true_value
row['meas_std'] = true_value * 0.2
data_table.append( copy.copy(row) )
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_rate_parent
'name': 'prior_rate_parent',
'density': 'uniform',
'lower': 1e-4,
'upper': 1.0,
'mean': 0.1,
},{ # prior_none
'name': 'prior_none',
'density': 'uniform',
'mean': 0.0,
}
]
# ----------------------------------------------------------------------
# smooth table
last_time_id = len(time_list) - 1
smooth_table = [
{ # smooth_rate_parent
'name': 'smooth_rate_parent',
'age_id': age_index,
'time_id': time_index,
'fun': fun_rate_parent
}
]
# ----------------------------------------------------------------------
# rate table
rate_table = [
{
'name': 'iota',
'parent_smooth': 'smooth_rate_parent',
},{
'name': 'rho',
'parent_smooth': 'smooth_rate_parent',
},{
'name': 'chi',
'parent_smooth': 'smooth_rate_parent',
}
]
# ----------------------------------------------------------------------
# option_table
option_table = [
{ 'name':'parent_node_name', 'value':'canada' },
{ 'name':'random_seed', 'value':random_seed },
{ 'name':'rate_case', 'value':'iota_pos_rho_pos' },
{ 'name':'quasi_fixed', 'value':'false' },
{ 'name':'derivative_test_fixed', 'value':'first-order' },
{ 'name':'max_num_iter_fixed', 'value':'30' },
{ 'name':'print_level_fixed', 'value':'5' },
{ 'name':'tolerance_fixed', 'value':'1e-12' },
{ 'name':'derivative_test_random', 'value':'second-order' },
{ 'name':'max_num_iter_random', 'value':'100' },
{ 'name':'print_level_random', 'value':'0' },
{ 'name':'tolerance_random', 'value':'1e-12' }
]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [ { 'subgroup':'world', 'group':'world' } ]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(
file_name,
age_list,
time_list,
integrand_table,
node_table,
subgroup_table,
weight_table,
covariate_table,
avgint_table,
data_table,
prior_table,
smooth_table,
nslist_table,
rate_table,
mulcov_table,
option_table
)
# ----------------------------------------------------------------------
n_smooth = len( smooth_table )
def example_db(file_name):
def fun_iota_parent(a, t):
return ('prior_iota_parent', 'prior_gauss_zero', 'prior_gauss_zero')
# ----------------------------------------------------------------------
# age_list
age_list = [0.0, 100.0]
#
# time_list
time_list = [1990.0, 2200.0]
#
# integrand table:
integrand_table = [
{
'name': 'Sincidence'
},
]
#
# node table:
node_table = [{'name': 'world', 'parent': ''}]
#
# weight table:
weight_table = list()
#
# covariate table:
covariate_table = list()
#
# mulcov table:
mulcov_table = list()
# -----------------------------------------------------------------------
# data table:
data_table = list()
#
# values that are the same for all data rows
row = {
'integrand': 'Sincidence',
'density': 'gaussian',
'weight': '',
'hold_out': False,
'age_lower': 50.0,
'age_upper': 50.0,
'time_lower': 2000.0,
'time_upper': 2000.0,
'node': 'world',
'subgroup': 'world',
}
# N = 2, y_0 = iota_true - noise , y1 = iota_true + noise
noise = 0.9 * iota_true
Sincidence = [
iota_true - noise,
iota_true + noise,
]
for meas_value in Sincidence:
row['meas_value'] = meas_value
row['meas_std'] = meas_value # 100 % coefficient of variation
data_table.append(copy.copy(row))
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_gauss_zero
'name': 'prior_gauss_zero',
'density': 'gaussian',
'mean': 0.0,
'std': 0.01,
},{ # prior_iota_parent
'name': 'prior_iota_parent',
'density': 'uniform',
'lower': 1e-19,
'upper': 1.0,
'mean': 1.01e-06,
}
]
#
# smooth table
name = 'smooth_iota_parent'
fun = fun_iota_parent
age_grid = [0]
time_grid = [0]
smooth_table = [{
'name': name,
'age_id': age_grid,
'time_id': time_grid,
'fun': fun
}]
#
# rate table:
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_iota_parent',
'child_smooth': None,
}]
#
# option_table
option_table = [
{
'name': 'rate_case',
'value': 'iota_pos_rho_zero'
},
{
'name': 'parent_node_name',
'value': 'world'
},
{
'name': 'quasi_fixed',
'value': 'false'
},
{
'name': 'print_level_fixed',
'value': '0'
},
{
'name': 'tolerance_fixed',
'value': '1e-9'
},
{
'name': 'tolerance_random',
'value': '1e-10'
},
]
#
# avgint table: empty
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
#
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
#
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
# ----------------------------------------------------------------------
return
def example_db(file_name):
def fun_rate_parent(a, t):
return ('prior_rate_parent', 'prior_gauss_zero', None)
# ----------------------------------------------------------------------
# age table
age_list = [0.0, 50.0, 100.0]
#
# time table
time_list = [1995.0, 2005.0, 2015.0]
#
# integrand table
integrand_table = [{'name': 'Sincidence'}]
#
# node table: world -> north_america
# north_america -> (united_states, canada)
node_table = [{
'name': 'world',
'parent': ''
}, {
'name': 'north_america',
'parent': 'world'
}, {
'name': 'united_states',
'parent': 'north_america'
}, {
'name': 'canada',
'parent': 'north_america'
}]
#
# weight table:
weight_table = list()
#
# covariate table: no covriates
covariate_table = list()
#
# mulcov table
mulcov_table = list()
#
# nslist_table:
nslist_table = dict()
#
# avgint_table
avgint_table = list()
# ----------------------------------------------------------------------
# data table: same order as age_list
data_table = list()
# values that are the same for all data rows
row = {
'node': 'canada',
'subgroup': 'world',
'density': 'gaussian',
'weight': '',
'hold_out': False,
'time_lower': 2000.0,
'time_upper': 2000.0,
'integrand': 'Sincidence',
'age_lower': 0.0
}
# values that change between rows: (one data point for each integrand)
for age_id in range(len(age_list)):
age = age_list[age_id]
meas_value = 1e-4 * (50.0 + age)
row['meas_value'] = meas_value
row['meas_std'] = 1e-4 * (50.0 + age_list[0])
row['age_lower'] = age
row['age_upper'] = age
data_table.append(copy.copy(row))
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_rate_parent
'name': 'prior_rate_parent',
'density': 'uniform',
'lower': 1e-4,
'upper': 1.0,
'mean': 0.01,
},{ # prior_gauss_zero
'name': 'prior_gauss_zero',
'density': 'gaussian',
'mean': 0.0,
'std': 1e-6,
}
]
# ----------------------------------------------------------------------
# smooth table
smooth_table = [{ # smooth_rate_parent
'name': 'smooth_rate_parent',
'age_id': range(len(age_list)),
'time_id': [0],
'fun': fun_rate_parent
}]
# ----------------------------------------------------------------------
# rate table
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_rate_parent',
}]
# ----------------------------------------------------------------------
# option_table: max_num_iter_fixed will be set later
option_table = [{
'name': 'parent_node_name',
'value': 'canada'
}, {
'name': 'ode_step_size',
'value': '10.0'
}, {
'name': 'random_seed',
'value': '0'
}, {
'name': 'rate_case',
'value': 'iota_pos_rho_zero'
}, {
'name': 'warn_on_stderr',
'value': 'false'
}, {
'name': 'quasi_fixed',
'value': 'true'
}, {
'name': 'derivative_test_fixed',
'value': 'first-order'
}, {
'name': 'print_level_fixed',
'value': '0'
}, {
'name': 'tolerance_fixed',
'value': '1e-12'
}, {
'name': 'derivative_test_random',
'value': 'second-order'
}, {
'name': 'max_num_iter_random',
'value': '100'
}, {
'name': 'print_level_random',
'value': '0'
}, {
'name': 'tolerance_random',
'value': '1e-10'
}]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
return
def example_db(file_name):
import copy
import dismod_at
import math
# ----------------------------------------------------------------------
# age table
age_list = [0.0, 100.0]
#
# time table
time_list = [1995.0, 2015.0]
#
# integrand table
integrand_table = [{'name': 'Sincidence', 'minimum_meas_cv': 0.0}]
#
# node table: world -> north_america
# north_america -> (united_states, canada)
node_table = [{
'name': 'world',
'parent': ''
}, {
'name': 'north_america',
'parent': 'world'
}, {
'name': 'united_states',
'parent': 'north_america'
}, {
'name': 'canada',
'parent': 'north_america'
}]
#
# weight table: The constant function 1.0 (one age and one time point)
fun = constant_weight_fun
weight_table = [{
'name': 'constant',
'age_id': [0],
'time_id': [0],
'fun': fun
}]
#
# covariate table:
covariate_table = list()
#
# mulcov table
mulcov_table = list()
# ----------------------------------------------------------------------
# data table:
data_table = list()
# values that are the same for all data rows
row = {
'density': 'gaussian',
'weight': 'constant',
'hold_out': False,
'time_lower': time_list[0],
'time_upper': time_list[-1],
'age_lower': age_list[0],
'age_upper': age_list[-1],
'meas_value': iota_true,
'meas_std': iota_true * 0.1,
'eta': None,
'integrand': 'Sincidence',
'subgroup': 'world',
}
# values that change between rows:
for data_id in range(n_data):
if data_id % 2 == 0:
row['node'] = 'north_america'
else:
row['node'] = 'canada'
data_table.append(copy.copy(row))
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_zero
'name': 'prior_zero',
'density': 'uniform',
'lower': 0.0,
'upper': 0.0,
'mean': 0.0,
'std': None,
'eta': None
},
{ # prior_none
'name': 'prior_none',
'density': 'uniform',
'lower': None,
'upper': None,
'mean': 0.0,
'std': None,
'eta': None
},
{ # prior_gauss_zero
'name': 'prior_gauss_zero',
'density': 'gaussian',
'lower': None,
'upper': None,
'mean': 0.0,
'std': 0.1,
'eta': None
},
{ # prior_iota_parent
'name': 'prior_iota_parent',
'density': 'gaussian',
'lower': iota_lower,
'upper': iota_upper,
'mean': iota_initial,
'std': 0.1,
'eta': 1e-6
}
]
# ----------------------------------------------------------------------
# smooth table
smooth_table = [
{ # smooth_rate_child
'name': 'smooth_rate_child',
'age_id': [0],
'time_id': [0],
'mulstd_value_prior_name': None,
'mulstd_dage_prior_name': None,
'mulstd_dtime_prior_name': None,
'fun': fun_rate_child
},
{ # smooth_rate_parent
'name': 'smooth_rate_parent',
'age_id': [0],
'time_id': [0],
'mulstd_value_prior_name': None,
'mulstd_dage_prior_name': None,
'mulstd_dtime_prior_name': None,
'fun': fun_iota_parent
},
{ # smooth_zero
'name': 'smooth_zero',
'age_id': [0],
'time_id': [0],
'mulstd_value_prior_name': None,
'mulstd_dage_prior_name': None,
'mulstd_dtime_prior_name': None,
'fun': fun_zero
}
]
# ----------------------------------------------------------------------
# rate table
rate_table = [{
'name': 'pini',
'parent_smooth': None,
'child_smooth': None,
'child_nslist': None
}, {
'name': 'iota',
'parent_smooth': 'smooth_rate_parent',
'child_smooth': 'smooth_rate_child',
'child_nslist': None
}, {
'name': 'rho',
'parent_smooth': None,
'child_smooth': None,
'child_nslist': None
}, {
'name': 'chi',
'parent_smooth': None,
'child_smooth': None,
'child_nslist': None
}, {
'name': 'omega',
'parent_smooth': None,
'child_smooth': None,
'child_nslist': None
}]
# ----------------------------------------------------------------------
# option_table
# use very large tolerance_fixed to avoid warning during test.
option_table = [{
'name': 'parent_node_name',
'value': 'north_america'
}, {
'name': 'rate_case',
'value': 'iota_pos_rho_zero'
}, {
'name': 'quasi_fixed',
'value': 'true'
}, {
'name': 'max_num_iter_fixed',
'value': '0'
}, {
'name': 'print_level_fixed',
'value': '0'
}, {
'name': 'tolerance_fixed',
'value': '1e+10'
}, {
'name': 'bound_frac_fixed',
'value': str(bound_frac)
}]
# ----------------------------------------------------------------------
# avgint table: empty
avgint_table = list()
# ----------------------------------------------------------------------
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
# ----------------------------------------------------------------------
n_smooth = len(smooth_table)
return
def example_db(file_name):
def fun_iota_n1(a, t):
return ('prior_iota_n1_value', None, None)
def fun_iota_child(a, t):
return ('prior_iota_child', None, None)
def fun_iota_subgroup(a, t):
return ('prior_iota_subgroup', None, None)
#
# node_table
node_table = [
{
'name': 'n1',
'parent': ''
},
{
'name': 'n11',
'parent': 'n1'
},
{
'name': 'n12',
'parent': 'n1'
},
{
'name': 'n111',
'parent': 'n11'
},
{
'name': 'n112',
'parent': 'n11'
},
{
'name': 'n121',
'parent': 'n12'
},
{
'name': 'n122',
'parent': 'n12'
},
]
# age_table
age_list = [0.0, 100.0]
# time_table
time_list = [1990.0, 2020.0]
# rate_table
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_iota_n1',
'child_smooth': 'smooth_iota_child',
}]
#
# subgroup_table
subgroup_table = [
{
'subgroup': 'none',
'group': 'none'
},
{
'subgroup': 'n111',
'group': 'n11'
},
{
'subgroup': 'n112',
'group': 'n11'
},
{
'subgroup': 'n121',
'group': 'n12'
},
{
'subgroup': 'n122',
'group': 'n12'
},
]
# covariate_table
covariate_table = [{'name': 'one', 'reference': 0.0}]
# mulcov_table
mulcov_table = [{
'covariate': 'one',
'type': 'rate_value',
'effected': 'iota',
'group': 'n11',
'smooth': None,
'subsmooth': 'smooth_iota_subgroup',
}, {
'covariate': 'one',
'type': 'rate_value',
'effected': 'iota',
'group': 'n12',
'smooth': None,
'subsmooth': 'smooth_iota_subgroup',
}]
# prior_table
prior_table = [
{ # prior_iota_n1_value
'name': 'prior_iota_n1_value',
'density': 'uniform',
'lower': iota_n1(0) / 100.0,
'upper': iota_n1(100) * 100.0,
'mean': iota_n1(50) / 2.0,
},{ # prior_iota_child
'name': 'prior_iota_child',
'density': 'uniform',
'mean': 0.0,
},{ # prior_iota_subgroup
'name': 'prior_iota_subgroup',
'density': 'uniform',
'mean': 0.0,
}
]
# smooth_table
smooth_table = [
{ # smooth_iota_n1
'name': 'smooth_iota_n1',
'age_id': [0, 1],
'time_id': [0],
'fun': fun_iota_n1
},
{ # smooth_iota_child
'name': 'smooth_iota_child',
'age_id': [0],
'time_id': [0],
'fun': fun_iota_child
},
{ # smooth_iota_subgroup
'name': 'smooth_iota_subgroup',
'age_id': [0],
'time_id': [0],
'fun': fun_iota_subgroup
}
]
# weight table:
weight_table = list()
# nslist_table
nslist_table = dict()
# option_table
option_table = [
{
'name': 'parent_node_name',
'value': 'n1'
},
{
'name': 'rate_case',
'value': 'iota_pos_rho_zero'
},
{
'name': 'zero_sum_child_rate',
'value': 'iota'
},
{
'name': 'quasi_fixed',
'value': 'false'
},
{
'name': 'max_num_iter_fixed',
'value': '100'
},
{
'name': 'print_level_fixed',
'value': '0'
},
{
'name': 'tolerance_fixed',
'value': '1e-12'
},
]
# integrand_table
integrand_table = [{'name': 'Sincidence'}]
# ------------------------------------------------------------------------
# data_table
data_table = list()
# values that are the same for all data points
row = {
'integrand': 'Sincidence',
'density': 'gaussian',
'weight': '',
'hold_out': False,
'time_lower': 2000.0,
'time_upper': 2000.0,
'one': 1.0,
}
random.seed(random_seed)
for age_id in range(len(age_list)):
age = age_list[age_id]
for node in ['n1', 'n11', 'n12', 'n111', 'n112', 'n121', 'n122']:
for i in range(data_per_node):
iota = avg_integrand(age, node)
meas_std = iota * meas_cv
meas_value = random.gauss(iota, meas_std)
if node in ['n1', 'n11', 'n12']:
row['subgroup'] = 'none'
else:
row['subgroup'] = node
row['node'] = node
row['meas_value'] = meas_value
row['meas_std'] = meas_std
row['age_lower'] = age
row['age_upper'] = age
data_table.append(copy.copy(row))
# ------------------------------------------------------------------------
# avgint_table
avgint_table = list()
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
def example_db (file_name) :
def constant_weight_fun(a, t) :
return 1.0
def fun_iota_parent(a, t) :
return ('prior_iota_parent', None, None)
# ----------------------------------------------------------------------
# age table
age_list = [ 0.0, 50.0, 100.0 ]
#
# time table
time_list = [ 1995.0, 2005.0, 2015.0 ]
#
# integrand table
integrand_table = [
{ 'name':'Sincidence' }
]
#
# node table: world -> north_america
# north_america -> (united_states, canada)
node_table = [
{ 'name':'world', 'parent':'' },
]
#
# weight table: The constant function 1.0 (one age and one time point)
fun = constant_weight_fun
weight_table = [
{ 'name':'constant', 'age_id':[1], 'time_id':[1], 'fun':fun }
]
#
# covariate table:
covariate_table = list()
#
# mulcov table
mulcov_table = list()
#
# avgint table: empty
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# data table:
data_table = list()
# Values that are the same for all data rows
# (because the simultion censors the data)
meas_std = iota_true;
eta = iota_true / 2.0 # large so get zeros for log scaled cases
row = {
'node': 'world',
'subgroup': 'world',
'integrand': 'Sincidence',
'density': density_name,
'weight': 'constant',
'hold_out': False,
'age_lower': 0.0,
'age_upper': 0.0,
'time_lower': 1995.0,
'time_upper': 1995.0,
'eta': eta
}
sigma = meas_std
mu = iota_true
if density_name in [ 'cen_log_gaussian', 'cen_log_laplace' ] :
sigma = math.log(iota_true + eta + meas_std)
sigma -= math.log(iota_true + eta)
mu = math.log(iota_true + eta )
scale = sigma / math.sqrt(2.0)
for data_id in range( n_data ) :
# simulate the data for distribution without censoring
if density_name in [ 'cen_laplace', 'cen_log_laplace' ] :
meas_value = numpy.random.laplace(mu, scale)
else :
meas_value = numpy.random.normal(mu, sigma)
# check for log transform case
if density_name in [ 'cen_log_gaussian', 'cen_log_laplace' ] :
meas_value = math.exp(meas_value) - eta
# censor the data; i.e., replace negative values by zero values.
meas_value = max(meas_value, 0.0)
row['meas_value'] = meas_value
#
delta = meas_std
if density_name in [ 'cen_log_gaussian', 'cen_log_laplace' ] :
# sigma = log(meas_value + eta + delta) - log(meas_value + eta)
delta = (math.exp(sigma) - 1.0) * (meas_value + eta)
row['meas_std'] = delta
#
data_table.append( copy.copy(row) )
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_iota_parent
'name': 'prior_iota_parent',
'density': 'uniform',
'lower': 0.001,
'upper': 1.00,
'mean': 0.1,
}
]
# ----------------------------------------------------------------------
# smooth table
smooth_table = [
{ # smooth_iota_parent
'name': 'smooth_iota_parent',
'age_id': [ 0 ],
'time_id': [ 0 ],
'fun': fun_iota_parent
}
]
# ----------------------------------------------------------------------
# rate table
rate_table = [
{
'name': 'iota',
'parent_smooth': 'smooth_iota_parent',
}
]
# ----------------------------------------------------------------------
# option_table
option_table = [
{ 'name':'random_seed', 'value':str(random_seed) },
{ 'name':'parent_node_name', 'value':'world' },
{ 'name':'rate_case', 'value':'iota_pos_rho_zero' },
{ 'name':'quasi_fixed', 'value':'false' },
{ 'name':'max_num_iter_fixed', 'value':'100' },
{ 'name':'print_level_fixed', 'value':'0' },
{ 'name':'tolerance_fixed', 'value':'1e-13' },
]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [ { 'subgroup':'world', 'group':'world' } ]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(
file_name,
age_list,
time_list,
integrand_table,
node_table,
subgroup_table,
weight_table,
covariate_table,
avgint_table,
data_table,
prior_table,
smooth_table,
nslist_table,
rate_table,
mulcov_table,
option_table
)
# ----------------------------------------------------------------------
return
def example_db(file_name):
def fun_iota_child(a, t):
return ('prior_iota_child', None, None)
def fun_iota_parent(a, t):
return ('prior_iota_parent', None, None)
# ----------------------------------------------------------------------
# age table
age_list = [0.0, 100.0]
#
# time table
time_list = [1995.0, 2015.0]
node_table = [{
'name': 'north_america',
'parent': ''
}, {
'name': 'united_states',
'parent': 'north_america'
}, {
'name': 'canada',
'parent': 'north_america'
}]
#
# weight table:
weight_table = list()
# integrand table
integrand_table = [{'name': 'Sincidence'}, {'name': 'susceptible'}]
#
# covariate table: no covriates
covariate_table = list()
#
# mulcov table
mulcov_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_iota_parent
'name': 'prior_iota_parent',
'density': 'uniform',
'lower': iota_north_america / 100.0,
'upper': iota_north_america * 100.0,
'mean': iota_north_america * 3.0,
},{ # prior_iota_child
'name': 'prior_iota_child',
'density': 'uniform',
'mean': 0.0,
}
]
# ----------------------------------------------------------------------
# smooth table
smooth_table = [
{ # smooth_iota_parent
'name': 'smooth_iota_parent',
'age_id': [0],
'time_id': [0],
'fun': fun_iota_parent
},
{ # smooth_iota_child
'name': 'smooth_iota_child',
'age_id': [0],
'time_id': [0],
'fun': fun_iota_child
}
]
# ----------------------------------------------------------------------
# rate table
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_iota_parent',
'child_smooth': 'smooth_iota_child',
}]
# --------------------------------------------------------------------
# option_table
option_table = [
{
'name': 'parent_node_name',
'value': 'north_america'
},
{
'name': 'ode_step_size',
'value': '1.0'
},
{
'name': 'rate_case',
'value': 'iota_pos_rho_zero'
},
{
'name': 'quasi_fixed',
'value': 'true'
},
{
'name': 'max_num_iter_fixed',
'value': '30'
},
{
'name': 'print_level_fixed',
'value': '0'
},
{
'name': 'tolerance_fixed',
'value': '1e-10'
},
]
# ----------------------------------------------------------------------
# data table:
data_table = list()
row = {
'density': 'log_gaussian',
'eta': '0.0',
'weight': '',
'hold_out': False,
'time_lower': 2000.0,
'time_upper': 2000.0,
'age_lower': 0.0,
'age_upper': 100.0,
'integrand': 'Sincidence',
'subgroup': 'world',
}
# north_america
row['node'] = 'north_america'
row['meas_value'] = iota_north_america
row['meas_std'] = row['meas_value'] / 10.0
data_table.append(copy.copy(row))
# canada
row['node'] = 'canada'
row['meas_value'] = math.exp(canada_effect) * iota_north_america
row['meas_std'] = row['meas_value'] / 10.0
data_table.append(copy.copy(row))
# united_states
row['node'] = 'united_states'
row['meas_value'] = math.exp(united_states_effect) * iota_north_america
row['meas_std'] = row['meas_value'] / 10.0
data_table.append(copy.copy(row))
# ----------------------------------------------------------------------
# avgint table:
avgint_table = list()
# values that are the same for all data rows
row = {
'integrand': 'susceptible',
'weight': '',
'time_lower': 2000.0,
'time_upper': 2000.0,
'age_lower': 50.0,
'age_upper': 50.0,
'subgroup': 'world'
}
row['node'] = 'north_america'
avgint_table.append(copy.copy(row))
row['node'] = 'canada'
avgint_table.append(copy.copy(row))
row['node'] = 'united_states'
avgint_table.append(copy.copy(row))
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)
def example_db(file_name):
def fun_rate_child(a, t):
return ('prior_gauss_zero', None, None)
def fun_rate_parent(a, t):
return ('prior_rate_parent', None, None)
# ----------------------------------------------------------------------
# age table
age_list = [0.0, 100.0]
#
# time table
time_list = [1995.0, 2015.0]
#
# integrand table
integrand_table = [{'name': 'Sincidence'}]
#
node_table = [
{
'name': 'world',
'parent': ''
},
{
'name': 'child_0',
'parent': 'world'
},
{
'name': 'child_1',
'parent': 'world'
},
]
#
# weight table:
weight_table = list()
#
# covariate table: no covriates
covariate_table = list()
#
# mulcov table
mulcov_table = list()
#
# avgint table:
avgint_table = list()
#
# nslist_table:
nslist_table = dict()
# ----------------------------------------------------------------------
# data table:
data_table = list()
row = {
'node': 'child_0',
'subgroup': 'world',
'density': 'gaussian',
'weight': '',
'hold_out': False,
'time_lower': 2000.0,
'time_upper': 2000.0,
'age_lower': 50.0,
'age_upper': 50.0,
'integrand': 'Sincidence',
'meas_value': y_0,
'meas_std': standard_dev,
}
data_table.append(copy.copy(row))
#
row['node'] = 'child_1'
row['meas_value'] = y_1
data_table.append(copy.copy(row))
#
# ----------------------------------------------------------------------
# prior_table
prior_table = [
{ # prior_rate_parent
'name': 'prior_rate_parent',
'density': 'uniform',
'lower': 1e-2 * theta_true,
'upper': 1e+2 * theta_true,
'mean': theta_true / 3.0, # only used for start and scale
},{ # prior_gauss_zero
'name': 'prior_gauss_zero',
'density': 'gaussian',
'mean': 0.0,
'std': standard_dev,
}
]
# ----------------------------------------------------------------------
# smooth table
smooth_table = [
{ # smooth_rate_parent
'name': 'smooth_rate_parent',
'age_id': [0],
'time_id': [0],
'fun': fun_rate_parent
},
{ # smooth_rate_child
'name': 'smooth_rate_child',
'age_id': [0],
'time_id': [0],
'fun': fun_rate_child
}
]
# ----------------------------------------------------------------------
# rate table
rate_table = [{
'name': 'iota',
'parent_smooth': 'smooth_rate_parent',
'child_smooth': 'smooth_rate_child',
}]
# ----------------------------------------------------------------------
# option_table
option_table = [{
'name': 'parent_node_name',
'value': 'world'
}, {
'name': 'ode_step_size',
'value': '10.0'
}, {
'name': 'random_seed',
'value': str(random_seed)
}, {
'name': 'rate_case',
'value': 'iota_pos_rho_zero'
}, {
'name': 'quasi_fixed',
'value': 'true'
}, {
'name': 'derivative_test_fixed',
'value': 'none'
}, {
'name': 'max_num_iter_fixed',
'value': '100'
}, {
'name': 'print_level_fixed',
'value': '0'
}, {
'name': 'tolerance_fixed',
'value': '1e-12'
}, {
'name': 'derivative_test_random',
'value': 'none'
}, {
'name': 'max_num_iter_random',
'value': '100'
}, {
'name': 'print_level_random',
'value': '0'
}, {
'name': 'tolerance_random',
'value': '1e-12'
}]
# ----------------------------------------------------------------------
# subgroup_table
subgroup_table = [{'subgroup': 'world', 'group': 'world'}]
# ----------------------------------------------------------------------
# create database
dismod_at.create_database(file_name, age_list, time_list, integrand_table,
node_table, subgroup_table, weight_table,
covariate_table, avgint_table, data_table,
prior_table, smooth_table, nslist_table,
rate_table, mulcov_table, option_table)